Lecture Notes: Epidemiology and Public Health Medicine

Lecture Notes

Epidemiology and Public Health Medicine

Richard Farmer
MB, PhD, FFPH, FFPM Professor of Epidemiology Postgraduate Medical School University of Surrey Stirling House Surrey Research Park Guildford Surrey, UK

Ross Lawrenson
MRCGP, FAFPHM, MD Dean of Medicine & Professor of Primary Health Care Postgraduate Medical School University of Surrey Stirling House Surrey Research Park Guildford Surrey, UK

Fifth Edition

 2004 by Blackwell Publishing Ltd Blackwell Publishing, Inc., 350 Main Street, Malden, Massachusetts 02148-5020, USA Blackwell Publishing Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK Blackwell Publishing Asia Pty Ltd, 550 Swanston Street, Carlton, Victoria 3053, Australia The right of the Authors to be identied as the Authors of this Work has been asserted in accordance with the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, except as permitted by the UK Copyright, Designs and Patents Act 1988, without the prior permission of the publisher. First published in 1977 under the title Lecture Notes on Epidemiology and Community Medicine Second edition 1983 Third edition 1991 Fourth edition 1996 Reprinited 1997, 1998 Fith edition 2004 Library of Congress Cataloging-in-Publication Data Farmer, R. D. T. Lecture notes on epidemiology and public health medicine / Richard D.T. Farmer, Ross Lawrenson. 5th ed. p. ; cm. Includes bibliographical references and index. ISBN 1-4051-0674-3 1. Epidemiology. 2. Public health. [DNLM: 1. Epidemiologic Methods. 2. Health Services. 3. Preventive Medicine. WA 950 F234L 2004] I. Title: Epidemiology and public health medicine. II. Lawrenson, Ross. III. Title. RA651.F375 2004 614.4 dc22 2004000864 ISBN 1-4051-0674-3 A catalogue record for this title is available from the British Library Set in 8/12 Stone Serif by SNP Best-set Typesetter Ltd., Hong Kong Printed and bound in India by Replika Press Pvt. Ltd. Commissioning Editor: Vicki Noyes Editorial Assistant: Nic Ulyatt Production Editor: Fiona Pattison Production Controller: Kate Charman For further information on Blackwell Publishing, visit our website: http://www.blackwellpublishing.com

Contents

Preface, vi List of Abbreviations, viii

Part 1 Epidemiology
1 General principles, 3 2 Cause and risk and types of epidemiological study, 7 3 Descriptive studies, 14 4 Surveys, survey methods and bias, 24 5 Cohort studies, 32 6 Casecontrol studies, 38 7 Intervention studies, 45 8 Health information and sources of data, 51 9 Indices of health and disease, and standardization of rates, 63 10 Medical demography, 69 11 Evidence-based medicine, 82

13 Health promotion and health education, 96 14 Control of infectious disease, 103 15 Immunization, 114 16 Environmental health, 127 17 Screening, 133

Part 3 Health Services

18 19 20 21 History and principles, 143 The National Health Service, 153 Health targets, 162 Evaluation of health services, 173

Appendices: Further Reading and Useful Websites

Appendix 1: Suggested further reading, 181 Appendix 2: Useful websites, 182 Index, 183

Part 2 Prevention and Control of Disease

12 General principles, 91

Preface

The UK Government is committed to improving the nations health and reducing health inequalities. Whilst the provision of health care is in a state of constant change it is important to remember that the key objective is to maintain and improve the health of the population. This was recognized by Derek Wanless in his report Securing Good Health for the Whole Population published on 25th February 2004. This document focused on prevention and the wider determinants of health. To prevent disease and improve health it is essential to understand why diseases arise; and conversely why, in many cases, they do not. To do this it is necessary to study the distribution and natural history of diseases in populations and to identify the agents responsible; effective strategies can then be planned. In the same way that the provision of health care should be evidence based, the introduction of new preventive strategies should be rigorously evaluated and researched. The application of evidencebased medicine is applicable to both clinical and public health practice. In the past the importance of public health medicine and the related basic medical sciences, in particular medical statistics and sociology applied to medicine, was not emphasized in the undergraduate medical education. This relative neglect changed in the 1990s with the GMC's recommendation on undergraduate medicine Tomorrows Doctors. This publication recommended that the theme of public health medicine should gure prominently in the undergraduate curriculum, encompassing health promotion and illness prevention, assessment and targeting of population needs and awareness of environmental and social factors in disease. This explicit and forceful advocacy for the discipline from a body as inuential as the GMC undoubtedly gave added momentum to the development of medical education. Similar changes emphasising the importance of disease prevention and the need to ensure that health care vi

is relevant effective and efcient are evident within the NHS in the UK as in many other countries. This is exemplied in the NHS plan The New NHS; modern, dependable (1997). This new edition of Lecture Notes: Epidemiology and Public Health Medicine, as before, covers the basic tools required for the practice of epidemiology and preventive health. The chapters in the rst section of the book outline the principles of epidemiology and lead the reader to some classic examples from the medical literature. A new chapter has been included on the practice of evidencebased medicine. The second section of the book covers the areas of prevention and control of disease in particular the chapter on health promotion has been updated to reect the advances that have occurred over the last eight years. The chapter on occupational health has been dropped from this edition. The nal section has been updated to reect the changes in the provision of health care. Change is now a constant in the health services and the shift between central control and devolution of responsibility will continue to ebb and ow. At the time of writing we are seeing more devolution of responsibility and the primary care trusts have a tremendous opportunity to deliver health services that are truly responsive to patient needs. We should also recognise the successes brought about through the introduction of health targets the incidence of heart disease is falling; the mortality from breast and cervical cancer has fallen as screening for these diseases has increased; and many infectious diseases, for practical purposes, have been eliminated. We still have many challenges obesity and diabetes are increasing rapidly, alcohol abuse has been recognized as a growing social problem and the spread of sexually transmitted disease and HIV still poses challenges. We hope readers will nd that this new edition continues to provide a basic structure to under-

Preface standing epidemiology and public health and that many of our readers will be encouraged to delve deeper into the subject. diseases and immunization. We must also recognise the contribution of Emeritus Professor David Miller who was the co-author of the rst four editions of this book. We would also like to thank Mrs Pat Robertson, our PA at the University, for her help and support. Richard Farmer Ross Lawrenson

Acknowledgements
We are greatly indebted to Dr Peter English of the Health Protection Agency for his help and support in the updating of the chapters on infectious

vii

List of Abbreviations

AHA AIDS BCG BMA CCDC CDSC CEHO CHAI DHA DoH DTP EBM FHSA GMC GPRD HEA HES Hib HIV HPA HSE ICD IHD IPV ITT MMR MRC NHS NHSME NICE OPCS OPV PCT PHLS PMR RAWP RCT RHA SARS SMR STD WHO viii

Area Health Authority acquired immune deciency syndrome bacille CalmetteGurin (vaccine) British Medical Association Consultant in Communicable Disease Control Communicable Disease Surveillance Centre Chief Environmental Health Ofcer Commission for Healthcare Audit and Inspection District Health Authority Department of Health diphtheria/tetanus/pertussis (vaccine) evidence-based medicine Family Health Service Authority General Medical Council General Practic Research Database Health Education Authority hospital episode statistics haemophilus inuenzae type b (vaccination) human immunodeciency virus Health Protection Agency Health and Safety Executive International Classication of Diseases ischaemic heart disease injected polio vaccine intention to treat measles/mumps/rubella (vaccine) Medical Research Council National Health Service National Health Service Management Executive National Institute for Clinical Excellence Ofce of Population Censuses and Surveys oral polio vaccine primary care trust Public Health Laboratory Service perinatal mortality rates Resource Allocation Working Party randomized controlled trial Regional Health Authority severe acute respiratory syndrome standardized mortality ratio sexually transmitted disease World Health Organization

Part 1 Epidemiology

Chapter 1 General principles

The word epidemiology is derived from Greek and literally means studies upon people. Modern methods of epidemiological enquiry were rst developed in the course of investigating outbreaks of infectious diseases in the 19th century. In contemporary medical practice the scope and applications of epidemiology have been greatly extended. Similar methods are now used in the investigation of the causes and natural history of all types of disease. They are also used in the development and assessment of preventive programmes and treatments, the assessment of the safety of medicines and in the planning and evaluation of health services. In contrast to clinical medicine, epidemiology involves the study of groups of people (populations) rather than the direct study of individuals. This does not diminish its relevance to clinical medicine. On the contrary, it enhances the practice of medicine by increasing the understanding of how diseases arise and how they might be managed both in the individual and in societies as a whole. Most doctors nd themselves involved with epidemiology through the use they make of the results of studies or sometimes as participants in investigations. It is important that all professionals involved in health care should have an understanding of the subject so that they can use epidemiological methods in the study of health and disease. More importantly, a knowledge of epi-

demiology is needed to appraise critically other peoples contributions.

The investigation of causes and natural history of disease

One of the most important roles of epidemiology is to provide a broader understanding of the causes and natural history of diseases than can be gained from the study of individuals. Clearly, the experience of an individual doctor is limited because the number of patients with a particular condition with whom he or she comes into contact is relatively small. The less frequent a disease, the more fragmentary is an individual doctors experience and understanding of it. If the experience of many doctors is recorded in a standard form and properly analysed then new and more reliable knowledge may often be acquired. This will assist in diagnosis, give a better understanding of prognosis and point to optimum management policies. Such systematic collection and analysis of data about medical conditions in populations is the essence of epidemiology. The value of pooling doctors experience in elucidating the causes of disease is well illustrated by the story of the epidemic of fetal limb malformations (phocomelia) that was caused by women taking the drug thalidomide during the rst trimester of pregnancy. Phocomelia, a major deformity in the development of the limbs, was a recognized 3

Chapter 1 General principles congenital abnormality long before the invention of thalidomide. A drawing by Goya called Mother with deformed child bears witness to the fact that it occurred in 18th century Spain (Fig. 1.1). Under normal circumstances it is a very rare abnormality. Any doctor may encounter such rare conditions at some time during his or her professional life. Little can be done to correct the malformation and, because the condition is well known, it is unlikely to warrant the preparation of a case report for publication. If, over a short period of time, each of a dozen or so doctors or midwives throughout the country delivered a child with such an abnormality, each would be personally interested but the signicance of these individual cases would pass unnoticed unless the doctors or midwives communicated with each other or there was a central reporting system. This is what happened early in the course of the thalidomide episode. One of the lessons learned was highlighted by the Chief Medical Ofcer in his 1966 annual report. He said that it . . . focused attention on the lack of information concerning the different types of congenital malformations. Had a national scheme for notication been available at this time, it is probable that the increase in limb deformities would have been noticed earlier and perhaps some of the tragedies could have been avoided. The thalidomide incident underlines the need to collect, collate and analyse data about the occurrence of disease in populations as a matter of routine. This will increase the probability that causes will be identied early and, whenever possible, eliminated. However, even with the most efcient and complete system of recording medical observations, it is unlikely that the causes of all disease will be identied. It is interesting to speculate about what would have happened had thalidomide been universally lethal to the fetus before the 12th week of pregnancy. The excess spontaneous abortions might have passed unnoticed, some even to the pregnant woman, and the possibility that thalidomide had any deleterious effect on the human fetus would not have come to light. The discovery of such causal relationships requires other approaches, but still depends on the study of populations and cannot be established by examination of individual cases. The same is true for most proposed causes (agents) and other factors which may determine or predispose to the occurrence of disease.

Disease in perspective
Another application of epidemiological techniques is to give perspective to the range of diseases facing doctors and the diversity of their natural history. The individual clinician only sees a selected and comparatively small proportion of sick people, and so may gain an erroneous impression of the relative frequency of different conditions in the community as a whole. He or she may also fail to appreciate the range of different ways in which diseases present and progress. This is important since, consciously or not, the clinician tends to rely on his or her personal experience to assess the likelihood of particular diagnoses and their prognosis when deciding management policy. Rather they should rely on unbiased evidence obtained from population studies.

Health care needs

Figure 1.1 Mother with deformed child by Francisco Jos Goya y Lucientes. (By courtesy of the Clich des Muses Nationaux, Paris.)

Apart from its signicance in day-to-day clinical practice, an unbalanced picture of disease inci-

General principles Chapter 1 dence or prevalence may also distort the view of the health care needs of the community. In the National Health Service and in most health care systems throughout the world, attempts are made to organize services according to priorities set by objective criteria rather than allowing them to be dictated solely by subjective judgements and traditional provision. An important report published in the early 1980s called Social Inequality and Health (The Black Report) drew attention to some of the major differences that persist in the patterns of illness and disability in England and in the use of health services between different socioeconomic groups. For example, men in social class V were reported to suffer from long-standing illnesses almost twice as often as those in social class I but they consulted their general practitioner only about 25% more often. This observation suggests a serious failure to match needs with appropriate services. It calls for detailed investigation of the relevant population groups to elucidate the reasons for it and the implications for future health care provision. services can be provided. Such trials are becoming increasingly numerous, but they usually need to be on a large scale to produce reliable results. Although this is expensive and time consuming it is necessary in the long-term interests of health care.

Clinical medicine and epidemiology

It will be clear from the above that there are important contrasts between the approaches to disease by clinicians and by epidemiologists. Recognition of these differences helps understanding of the subject. The clinician asks the question What disease has my patient got? whereas the epidemiologist asks Why has this person rather than another developed the disease? How could it be prevented? Why does the disease occur in winter rather than summer? Why in this country but not in another? In order to answer such questions it is necessary to compare groups of people, looking for factors that distinguish people with disease from those without. Underlying the investigation of disease in this way is the belief that the misfortune of an individual in contracting a disease is not due to chance or fate but to a specic, denable and preventable combination of circumstances or personal characteristics. For a clinician, the utility of a diagnosis is a pointer to management decisions. Therefore the diagnostic precision required is related to the specicity of treatments that are available. For an epidemiologist, diagnosis has different signicance. It is a way of classifying individuals in order to make comparisons between groups. Lack of diagnostic precision will result in poor denitions of categories. This makes it difcult to identify the subtle yet important differences between groups which are critical to the understanding of the causes and prevention of disease. The clinician is interested in the natural history of disease for prognostic purposes in an individual patient. He or she is usually content to express prognosis in terms such as good, bad, about 6 months, etc. It is unhelpful to the clinician and the patient to attempt to introduce mathematical precision into prognostic statements, such as He 5

Evaluation of medical interventions

Epidemiology is of value in testing the usefulness (and safety) of medical interventions. Although many existing remedies have never been subjected to trial, everyone nowadays recognizes the necessity to conduct clinical trials of a new drug or vaccine before it is introduced into medical practice. This is the only way to demonstrate that a particular drug or vaccine is likely to improve the patients prospects of recovery or to prevent disease from occurring or progressing. Once a product has been launched on the market it is necessary to continue to monitor its effects (both benecial and adverse) in order to ensure that patients are being prescribed effective and safe medication. In recent years the application of epidemiological methods to the assessment of medicines has become rmly established and is referred to as pharmacoepidemiology. The same principles are being applied to other treatments, such as surgery or physical therapy, and even to the alternative ways in which health

Chapter 1 General principles has a 10.9% chance of surviving symptom-free for 5 years, though it may sometimes be appropriate to give a range of expected survival times, for example between 3 and 7 years. By contrast, in population studies precision is helpful because it may allow the investigator to identify variables that have signicant effects on outcome. For example, it may be informative to investigate why in one group of patients 10.9% survive symptom-free for 5 years while in another group with approximately similar conditions, 26.5% survive symptom-free for 5 years. What accounts for this difference which could assist in planning treatment or preventive strategies? While there are these clear differences between clinical and epidemiological approaches to medical problems and while their immediate purposes are different, it is also clear that the results of epidemiological investigations can contribute greatly to the scientic basis of clinical practice.

Chapter 2 Cause and risk, and types of epidemiological study

Introduction
The principal uses of epidemiology in medicine have been described in Chapter 1. These are: the investigation of the causes and natural history of disease, with the aim of disease prevention and health promotion; and the measurement of health care needs and the evaluation of clinical management, with the aim of improving the effectiveness and efciency of health care provision. Both involve the important and fundamental concepts of cause and risk. The concept of cause must be distinguished from the notion of association. Not all factors that are statistically associated with the occurrence of disease are causes. They also include so-called determinants, confounding variables and factors associated by pure chance.

Concept of cause
A cause is an event, characteristic or condition that precedes the disease and without which the disease could not have occurred. The event may be exposure to a microbe, chemical substance, physical trauma, radiation or other exposure. Many diseases do not have a single cause and thus exposure to a causal agent does not inevitably result in disease. For example, smoking tobacco is a cause of lung cancer; however, not all individuals who smoke will develop lung cancer. Those who do will

have other exposures or characteristics that act with the effects of tobacco smoking to cause the disease. Venous thrombosis is caused by a combination of stasis, vessel wall damage and a hypercoagulable state (Virchows triad). An individual may have a disorder that results in a hypercoagulable state (for example, inherited disorders of the coagulation system such as factor V Leiden) yet never have a venous thrombosis because he or she never experiences the concurrence of vessel wall trauma and stasis necessary to produce the disease. Thus, the risk of deep venous thrombosis in such individuals is measurably increased but it is not inevitable. Although the cause of a disease is always statistically associated with its occurrence a statistical association cannot be taken as proof of cause. Sometimes an event or exposure is associated with both the occurrence of the disease and another exposure which is statistically associated with the disease. This is called confounding. For example, if one were to investigate the association between alcohol consumption and coronary heart disease, smoking would be a confounding exposure because smoking tends to be positively associated with alcohol consumption and is also a cause of coronary heart disease. If the presence of confounding is not allowed for in such a study then it might result in the misleading conclusion that alcohol is directly associated with coronary heart disease. 7

Chapter 2 Cause and risk Statistically signicant associations between exposure and the occurrence of disease may occur by chance, i.e. they are neither causal factors nor confounding factors. A determinant is an attribute or circumstance that affects the liability of an individual to be exposed to or, when exposed, to develop disease (e.g. hereditary predisposition, environmental conditions). A confounding variable is a factor that is signicantly associated both with the occurrence of a disease in a population and with one of its causes or determinants, but is not itself a cause. For example, heavy cigarette smoking and a high alcohol consumption tend to occur together. Smoking is causally associated with carcinoma of the bronchus and because heavy drinking is associated with cigarette smoking, alcohol consumption will tend to correlate with carcinoma of the bronchus, even though it is not a cause. The concept of risk includes both the risk that a person exposed to a potentially harmful agent will develop a particular disease and the risk that a particular intervention will benecially or adversely inuence the outcome. The indices commonly used to measure risk are set out below. Risk factors are different but are involved in both concepts. They are factors that are associated with a particular disease or outcome. They can be associated either by chance or because they inuence the course of events. All causal agents and determinants are risk factors but not all risk factors are causal agents or determinants. The purpose of epidemiological studies is to identify causes and determinants and to dene and measure risks by the application of the scientic methods set out in the next four chapters. to the direct cause. Thus, the investigation of cause is usually a complex exercise that involves the identication of both the characteristics of susceptible individuals (and sometimes characteristics of individuals who appear to be unusually resistant) and the types of exposure to external agents that are necessary for the disease to occur. Ideally, causal hypotheses should be explored by carefully controlled experiments in which the effects of each of the postulated causes can be examined independently of other factors. In animal studies, for example, it is usually possible to exclude the effects of inheritance by breeding a family of animals for study. The possible effects of the general environment and diet that are not of interest for a particular investigation can be eliminated by rearing the whole family under standard conditions. Then the effects of a suspected causal agent can be assessed by exposing a sample of the animals to it whilst protecting others from it. In such experiments the only major difference between the two groups is their exposure to the agent under study. Such a study design allows the observed effects, if any, to be attributed unequivocally to the agent under investigation. It is impractical and unethical to undertake studies of such experimental purity amongst human subjects. The identication of the causes of diseases and factors that alter the course of a disease in humans necessitates adopting methods whereby hypotheses can be tested without prejudice to the individuals being studied. The methods that are used in epidemiological studies represent practical compromises of the above ideal design. It is essential therefore that the results of any investigation are interpreted in full knowledge of the limitations imposed by the compromises. In particular, it is important to take account of the effects of confounding variables and, when these cannot be controlled in the study design, to allow for them in the analysis.

Causes and determinants

Few diseases have a single cause. Most are the result of exposure of susceptible individuals to one or more causal agents. Even in the case of some of the most straightforward illnesses, for example infections, exposure to the causal agent does not inevitably result in disease. Many other factors may inuence the development of disease in addition 8

Distinguishing causes and determinants from chance association

The observation that a disease is statistically associated with a suspected agent is clearly not proof that

Cause and risk Chapter 2 the suspected agent causes the disease. For example, there is a higher prevalence of alcoholism amongst publicans and bar staff than in most other occupational groups. This does not necessarily mean that being a publican causes alcoholism. There are several other possible explanations of this phenomenon, including the fact that people who tend towards excessive alcohol consumption may seek jobs in bars. The types of evidence that can be used to distinguish a causal from a fortuitous association are discussed below. Many of the criteria appear to be simple and straightforward but it can be seen that each of them can present practical difculties. (HIV). Most people with AIDS could have become infected with HIV on many occasions. By the time the disease is apparent it is impossible to prove that a particular exposure or type of activity led to the infection. In some circumstances it is not possible to date the start of the disease; for example, carcinoma of the endometrium usually occurs many years before symptoms are manifest and the disease is diagnosed. In such cases, although it is usually possible to date exposures to suspected causal agents they cannot be related in time to the disease.

Distribution of the disease

The spatial or geographical distribution of the disease should be similar to that of the suspected causal agent. For example, endemic goitre occurs in areas where the iodine content of drinking water is low. Sometimes a geographical association between the distribution of the disease and its suspected causal agent may be difcult to demonstrate. This is a particular problem if there is a signicant time interval between exposure and manifestation of disease and there have been movements in the population during that interval. For example, legionnaires disease commonly occurs in people who become infected as a result of casual or transient exposure to the source and who may be widely scattered before they develop symptoms of the disease. In these circumstances it is necessary to map the location of cases to the place where they were at the time it is hypothesized that they were exposed to the causal agent.

Distinguishing cause from association

Strength of association Time sequence Distribution of the disease Gradient Consistency Specicity Biological plausibility Experimental models Preventive trials

Strength of association
The stronger the association the more likely it is to be causal. This is usually measured in terms of relative risk, i.e. the incidence of disease in people exposed to the suspected agent compared with the incidence in those not so exposed (see below).

Gradient
The incidence of disease should correlate with the amount and duration of exposure to the suspected cause (population doseresponse). For example, mesothelioma was noted to be more common than expected in people working with asbestos and in those living near to factories that emitted asbestos dust into the atmosphere. The incidence was greatest in workers exposed for the longest periods and those living in closest proximity to the factories. 9

Time sequence
If an agent causes a disease then exposure must always precede its onset. Thus eating contaminated food can cause diarrhoea and vomiting 24 h later. A practical problem is that it is often difcult to date exposure to a suspected causal agent; for example, the acquired immune deciency syndrome (AIDS) is usually not manifest until many years after infection with the human immunodeciency virus

Chapter 2 Cause and risk a hypothesis. For example, in the mid-19th century, John Snow suggested that cholera was caused by an invisible agent in water. The epidemiological data were entirely consistent with the hypothesis but the cholera vibrio and its mode of spread had yet to be discovered.

Consistency
The same association between a disease and a suspected causal agent should be found in studies of different populations. Failure to nd consistency may be explained by differences in study design. Caution is needed before rejecting a causal hypothesis in such circumstances. For example, studies designed to test the hypothesis that carcinoma of the breast is causally associated with exposure to oral contraceptives have produced conicting results. Some appear to demonstrate that women exposed to oral contraceptives over long periods of time have an increased risk of breast cancer; others do not support this hypothesis. Careful review of the studies reveals differences in the criteria for the selection of cases and in the analytic techniques used, which may explain the apparently conicting results. A causal hypothesis can be regarded as supported only when there is a general consistency of ndings from studies conducted in the same way.

Experimental models
The disease can be reproduced in experimental models with animals. The fact that exposure to an agent can produce a disease in animals similar to that seen in humans gives credence to a causal hypothesis. However, failure to produce the disease amongst animals cannot be used as evidence to reject the hypothesis. For example, some microorganisms are pathogenic in humans but not usually in animals (e.g. measles virus); others are pathogenic in animals but not usually in humans, and only a minority are normally pathogenic in both.

Preventive trials Specicity

Specicity was amongst the criteria that could be used to distinguish chance associations from cause suggested by Hill in 1965. He proposed that a single true cause should lead to a single effect, not multiple effects. This criterion is particularly useful for infectious agents. It is not necessarily valid for non-infectious disease since it is widely accepted that a single agent can be causally associated with a number of outcomes; for example smoking cigarettes can cause lung cancer, heart disease and chronic obstructive airway disease, amongst other diseases. Control or removal of the suspected agent results in decreased incidence of disease. For example, when it was appreciated that the use of thalidomide for treatment of morning sickness in pregnancy was associated with a high incidence of phocomelia, the drug was withdrawn and the epidemic rapidly ceased.

Risk
There are three common indices of risk: absolute, relative and attributable.

Biological plausibility
The association between the disease and exposure to the suspected causal agent should be consistent with the known biological activity of the suspected agent. Sometimes an association is observed before the biological process is identied. The fact that there is no known biological explanation for an association should not on its own lead to rejection of 10

Types of risk
Absolute: incidence of disease in any dened population Relative: ratio of the incidence rate in the exposed group to the incidence rate in the non-exposed group Attributable: difference between the incidence rates in the exposed and non-exposed groups

Cause and risk Chapter 2 Absolute risk in cigarette smokers = 5.16 per 1000 Absolute risk in non-smokers = 0.55 per 1000 Relative risk in cigarette smokers = 5.16/0.55 = 9.38 Attributable risk of cigarette smoking = 5.16 0.55 = 4.61 per 1000 This indicates that smokers were 9.38 times more likely to die during the 7-year period than nonsmokers and that the additional risk of death carried by smokers compared with non-smokers was 4.61 per 1000 people per 7 years. The condence with which these ndings can be applied to the general population is determined in part by the similarity of the two groups in respect of attributes other than their smoking habits, in part upon whether the smokers are representative of the whole population of smokers and in part upon the sizes of the samples investigated. If the sampling was truly representative, the proportion of deaths in smokers that would be eliminated by cessation of smoking is the ratio of attributable to absolute risk (4.61/5.16 = 89%). This is known as the attributable fraction.

Absolute risk
This is the most basic measurement; it is the incidence of a disease in any dened population. The denominator can be the whole population or a subpopulation dened on the basis of an exposure. The absolute risk in an exposed population taken in isolation is often not a very useful index. To be meaningful it has to be compared with the risk in an unexposed population.

Relative risk
This is the ratio of the incidence rate in the exposed group to the incidence rate in the non-exposed group. It is a measure of the proportionate increase (or, if the agent is protective, the decrease) in disease rates of the exposed group. Thus, it makes allowance for the frequency of the disease amongst people who are not exposed to the supposed harmful agent. It is important to consider the relative risk in conjunction with the absolute risk. For example, a relative risk of 3 (people exposed have three times the risk of those not exposed) can cause concern. However, if the absolute risk is 1 in 100 000 it is less worrying than if the risk is 1 in 100.

Types of epidemiological study

There are four broad types of epidemiological study: descriptive cohort casecontrol intervention. They serve different purposes. None of them is entirely clear cut and it is not protable to try to classify each and every study within these classical types. Frequently the detailed investigation of a disease involves undertaking several studies of different types. They are dened and explained here to enable the reader to understand the concepts involved and to provide a framework which can be used to identify the most appropriate study design to answer particular problems. They are discussed in greater detail, with examples, in ensuing chapters.

Attributable risk
This is the difference between the incidence rates in the exposed and the non-exposed groups, i.e. it represents the risk attributable to the factor being investigated. The use of these measures of risk can be illustrated with data collected during the course of a cohort study which compared mortality amongst cigarette smokers with non-smokers during a 7-year period (Table 2.1).
Table 2.1 A comparison of mortality amongst cigarette smokers and non-smokers. Death rate Number in Died within over 7 years study 7 years (per 1000) Cigarette 25 769 smokers Non-smokers 5 439 133 3 5.16 0.55

11

Chapter 2 Cause and risk studies rely on data collected in a systematic manner according to well-dened procedures. In a cohort investigation individuals are selected for study on the basis that they are or may be exposed to the agent under investigation and are readily identied and followed up for a period of time. The follow-up may extend into years and aims to identify the characteristics of those who develop the disease (or other prior dened end point) and those who do not. The subjects investigated in a casecontrol study are generally recruited because they already have the disease (or end point) being investigated. Their past histories of exposure to suspected causal agents are compared with those of control subjectsindividuals who are not affected with the disease but are drawn from the same general population. The analysis involves discriminating between the past exposures and other relevant characteristics of the cases and those of the controls. The differences between these two study designs are schematically represented in Fig. 2.1. The cohort study design is closest to the ideal experimental design. Such studies tend to take longer and to be more expensive than casecontrol studies. However, they usually yield more robust ndings. Casecontrol studies, though usually cheaper

Descriptive studies
These are used to demonstrate the patterns in which diseases and associated factors are distributed in populations. They aim to identify changes in mortality and morbidity in time or to compare the incidence or prevalence of disease in different regions or between groups of individuals with different characteristics (e.g. occupational groups). Correlations are then sought with one or more other factors which may be thought to inuence the occurrence of the diseases. Studies of this type may give rise to hypotheses of cause but cannot be used in isolation to explore the meaning of associations and can rarely prove cause. This requires the use of the other types of study.

Cohort and casecontrol studies

These studies are observational. They are planned investigations designed to test specic hypotheses. They aim to dene the causes or determinants of diseases more precisely than is possible using descriptive studies alone. They do not involve the investigator in determining the exposures of individuals. From the results, it is often possible to suggest ways whereby the disease may be prevented or controlled. Both cohort and casecontrol

Past history risk factors Cohort

Composition of study population Exposed/at risk Whole population or random sample Not exposed/at risk

Future disease Disease No disease Disease No disease

Casecontrol

Present Absent Present Absent Figure 2.1 Comparison of cohort and casecontrol study designs. Cases of disease Matched controls

12

Cause and risk Chapter 2 and quicker to complete than cohort studies, rarely give clear-cut proof of cause. ies of treatment, prevention and control measures and the way in which health care is provided. They can also be used to assess the comparative effectiveness and efciency of different interventions. The most familiar study design of this type is the clinical trial. Ethical considerations are particularly important when considering the design and execution of any kind of intervention study.

Intervention studies
These are essentially experiments designed to measure the efcacy and safety of particular types of health care intervention. This can include stud-

13

Chapter 3 Descriptive studies

Introduction
Often an important starting point for many epidemiological investigations is the description of the distribution of disease in populations (descriptive studies). The principal advantages of descriptive studies are that they are cheap and quick to complete and they give a useful initial overview of a problem that may point to the next step in its investigation. Usually, descriptive studies make use of routinely collected health data, for example death certication data, hospital admission statistics, collated data from computerized general practices or infectious disease notications. The main sources of routine health data are set out in Chapter 8. Some social and other variables in relation to which disease data may be examined are also available from a wide variety of routine sources. The actual source used for a particular investigation depends on the data that are required. With the exception of census material, routine sources of social data are not discussed in detail in this book. Often the data required to describe disease distribution in a population and related variables are not readily available or are unsatisfactory for epidemiological purposes. In these circumstances it is necessary to collect the raw material for a descriptive study by special surveys. These surveys are usually cross-sectional in type (see Chapter 4).

Use of descriptive studies

Aetiological
The results of descriptive studies usually only give general guidance as to possible causes or determinants of disease, for example where broad geographical differences in prevalence are shown. Sometimes they may be quite precise, for example where a particular disease is very much more frequent within an occupational group or only occurs in a particular exposure group (e.g. asbestosis). Analysis of the data may indicate that certain attributes or exposures are more commonly found amongst people who have the disease than in those who do not. The converse may also be demonstrated, namely that certain attributes are more commonly found amongst people who do not have the disease than in those who do. This may be an equally valuable nding. It is rarely possible to prove that an agent causes a disease from a descriptive study, but investigations of this type will often generate or support hypotheses of aetiology and justify further investigations.

Clinical
Clinical impressions of the frequency of different conditions and their natural history are often misleading. The clinical impression is inuenced by the special interests of individual doctors, by

14

Descriptive studies Chapter 3 events that make a particular impression and by the chance clustering of cases. To obtain a balanced view of the relative importance of different conditions, their natural history and the factors that affect outcome requires data from a total population or an unbiased (random) sample. Knowledge of the relative frequency of different diseases is helpful to the clinician when deciding on the most likely diagnosis in individual patients. The probabilities of different diseases vary at different times and in different situations.

Variation of disease with time

Long-term (secular) trends Periodic changes (including seasonality) Epidemics

Long-term (secular) trends

These are changes in the incidence of disease over a number of years that do not conform to an identiable cyclical pattern. For example, the secular trend in mortality from tuberculosis in England and Wales has showed a steady fall over many years (Fig. 3.1) but recently the annual number of cases has started to rise. The observation of this trend on its own does not give any indication of its cause. However, it is sufciently striking to justify specic studies aimed at trying to identify the reasons for the change. The inclusion in Fig. 3.1 of the times at which various discoveries were made or specic measures were introduced gives some enlightenment. The overall trend seems to have been hardly affected by the identication of the causal organism, or by the introduction of chemotherapy and bacille CalmetteGurin (BCG) vaccination. This suggests that these played little part in the decline in mortality. However, the presentation of these data on an arithmetic scale (as in Fig. 3.1) disguises an important feature of the trend, i.e. a change in the rate at which the decline occurred. When the

Service planning
Health service planning in the past has been largely based on historical levels of provision and responses to demands for medical care. In order to plan services to meet needs rather than demand, and to allocate resources appropriately, accurate descriptive data are required on the relative importance and magnitude of different health problems in various segments of the community. They are also essential in order to evaluate the effectiveness of services and to monitor changes in disease incidence which may indicate a need for control action or the reallocation of resources and adjustments to service provision.

Analysis of descriptive data

Data derived from routine mortality and morbidity statistics (and from cross-sectional surveys) are usually analysed within three main categories of variable: time (when?) place (where?) personal characteristics (who?).

1600 1400 SMR (base years 195052) 1200 1000 800 600 400 200 1855 1875 1895 1915 Years 1935 1955 1965 Chemotherapy and BCG vaccination Organism discovered

Time
Three broad patterns of variation of disease incidence with time are recognizable. These are shown below.

Figure 3.1 Tuberculosis mortality in England and Wales, 18551965 (arithmetic scale).

15

Chapter 3 Descriptive studies data are plotted on a logarithmic scale (Fig. 3.2) it becomes clear that the introduction of specic measures for the control and treatment of tuberculosis was associated with an acceleration in the established decline in mortality. It is now thought that the decline in mortality from tuberculosis was due to a complex series of changes. Until the 1950s, these were mainly an increase in the resistance of the population to infection and environmental changes that reduced the chances of acquiring infection. After the early 1950s, the rate of decline in mortality was accelerated by the newly available methods of management. It is frequently necessary to examine secular trends both as changes in rates (arithmetic scale) and as rates of change (logarithmic scale) if the nature of a trend is to be fully appreciated. The secular trend in mortality from carcinoma of the bronchus shows the opposite picture to that for tuberculosis (Fig. 3.3). Until quite recently it had been increasing relentlessly amongst males but the rate of increase has now declined. By contrast, the increase in mortality rates amongst women continues. The powerful correlation between mortality and changes in the national consumption of cigarettes gave rise to the hypothesis that cigarette smoking could be the causal agent, although it did not prove causality. The hypothesis has since been explored through large numbers of epidemiological studies.

2000

1000

SMR (base years 195052)

Mass radiography BCG vaccine Chemotherapy 100

10 1871 1891 1911 1931 Years 1951 1971

Periodic changes
These are more or less regular or cyclic changes in incidence. The most common examples are seen in infectious diseases. For example, until a vaccine was introduced, measles showed a regular biennial cycle in incidence in England and Wales (Fig. 3.4). The cycles were probably the result of changes in

Figure 3.2 Tuberculosis mortality in England and Wales, 18711971 (logarithmic scale). (Reproduced with permission from Prevention and Health: Everybodys Business, HMSO, 1976.)

1200 1000 Deaths per million 800 600 400 200 0 1955 1960 1965 1970
Cigarett

Male (deaths)

140 Cigarette consumption 109 120 100

e consum

ption

80

(de Female

aths)

60 40 20

1975 Year

1980

1985

1990

Figure 3.3 Carcinoma of lung, bronchus and trachea. Deaths per million population in England and Wales, 195592, and cigarette consumption per year. (Reproduced with permission of the Ofce of National Statistics).

16

Descriptive studies Chapter 3

800 000

600 000 Notifications Figure 3.4 Notications of measles in England and Wales showing periodic variation (prior to introduction of measles vaccination). (Reproduced with permission of the Ofce of Population Censuses and Surveys (Crown copyright).)

400 000

200 000

42 44 46 48 50 52 54 56 58 60 62 64 66 68 Years (194268)

the levels of child population (herd) immunity (see p. 105). Other infectious diseases such as whooping cough, rubella and infectious hepatitis show less regular, but nevertheless distinct, cycles with longer intervals between peaks.

Seasonality This is a special example of periodic change. The environmental conditions that favour the presence of an agent, and the likelihood of its successful transmission, change with the seasons of the year. Respiratory infections, which spread directly from person to person by the air-borne route, are more common in winter months when people live in much closer contact with each other than in the summer. By contrast, gastrointestinal infections, which spread by the faecaloral route, often through contamination of food, are more com-

mon in summer months when the ambient temperatures favour the multiplication of bacteria in food. The regular seasonality of gastrointestinal infections is shown in Fig. 3.5 in which the number of notications of food poisoning for each quarter in 197489 are plotted. A particularly interesting feature of food poisoning incidence is that the marked seasonality is combined with a noticeable secular trend. The number of cases notied from late 1988 and early 1989 was much higher than in previous years. This could be due to contamination in the food chain, a decline in standards of food storage, distribution or preparation, or the result of an increase in notication rates following publicity given to the problem of food poisoning. Some non-infectious conditions, for example allergic rhinitis, deaths from drowning and road accidents, also display distinct seasonality. For

16 000 14 000 12 000 Cases notified 10 000 8 000 6 000 4 000 2 000 Figure 3.5 Quarterly notications of food poisoning in England and Wales, 197489. 0 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 Years (197489)

17

Chapter 3 Descriptive studies

Table 3.1 Seasonality of birth of schizophrenic and neurotic people compared with that of the general population (expected) showing an increased frequency of births of schizophrenic people in the rst part of the year but no seasonality amongst neurotic people. (Adapted from Hare E, Price J, Slater E. Br J Psychiatry 1974; 124: 8186.) Quarter of birth JanMar Schizophrenic people Observed Expected Observed as a percentage of expected Neurotic people Observed Expected Observed as a percentage of expected AprJune JulySept OctDec

1383.0 1292.1 107

1412.0 1342.8 105

1178.0 1293.1 91

1166.0 1211.1 96

3085.0 3024.1 101

3172.0 3150.6 101

2949.0 3042.0 97

2882.0 2844.2 101

most of these, the explanation for the seasonality is not difcult to infer. There are seasonal variations in the incidence of certain other conditions, however, for which there is as yet no rational explanation. For example, schizophrenic people are more likely than the general population to be born in the early months of the year (February and March) (Table 3.1). Many hypotheses have been offered to explain this observation, including the proposition that the disease is caused by an intrauterine infection, that the mothers of schizophrenic people are more likely to miscarry at certain times of the year (thereby resulting in a decit of births in months other than January to March) and that the mothers are more likely to conceive in April to June than are other women. However, none has yet been proved.

It should be noted that the seasonality in disease patterns related to climatic conditions is reversed in the southern Hemisphere.

Epidemics
These are temporary increases in the incidence of disease in populations. The most obvious epidemics are of infectious diseases such as inuenza (Fig. 3.6) but non-infectious epidemics do occur. For example, there was an increase in asthma deaths in the 1960s associated with the increased use of pressurized aerosol bronchodilators (Fig. 3.7). The word epidemic is also sometimes used to describe an increase in incidence above the level expected from past experience in the same population (or from experience in another population

1400 1200 Number of deaths 1000 800 600 400 200 0 1 14 27 40 1 14 27 Week 40 1 14 27 40 Figure 3.6 Weekly deaths from inuenza in England and Wales, 197577.

18

Descriptive studies Chapter 3

800 700 Sales and prescriptions (thousands) 600 500 400 300 200 100 0 40 30 20 10 0 Direct sales Prescriptions Total sales

Figure 3.7 Sales and prescriptions of asthma preparations compared with deaths from asthma among people aged 534 years, in England and Wales, 195968. (After Inman WHW, Adelstein AM. Lancet 1969; ii: 279.)

Deaths

59

60

61

62

63 64 65 Years (195968)

66

67

68

with similar demographic and social characteristics). However, if the strict denition of epidemic is used, it is inappropriate to use the term to describe secular trends in diseases such as diabetes or malignant melanoma, since there is no evidence that they are temporary increases in incidence.

Broad geographical differences

Variations in the incidence of disease are sometimes related to factors such as climate, social and cultural habits (including diet), and the presence of vectors or of other naturally occurring hazards. Although the incidence of disease does not respect administrative boundaries between countries or regions, these boundaries often follow broadly natural ecological boundaries and tend to encompass common social and cultural groups. Much valuable information pointing to possible causes of disease has been obtained by comparisons of routinely collected data between countries and other administrative units. For example, various forms of cancer and other conditions show striking geographical difference in incidence (Table 3.2).

Place
Variations in the incidence or prevalence of disease with place can be considered under three headings.

Variation of disease with place

Broad geographical differences Local differences Variations within single institutions

Local differences
The distribution of a disease may be limited by the localization of its cause. Thus, if a main water sup19

Chapter 3 Descriptive studies

Table 3.2 Geographical variation in the incidence of disease. Comparison of death rates in England and Wales with those in Japan (1979) for various causes shows considerable discrepancies. Both are highly industrialized countries with welldeveloped health services, but they have very different cultures and racial origins. (Data from World Health Statistics Annual, WHO, Geneva, 1981.) Rates per 100 000 Disease England and Wales high, Japan low Cancer of breast (females) Cancer of prostate Cancer of colon England and Wales low, Japan high Cancer of stomach Cirrhosis of liver (males) Suicide England and Wales Japan

47.9 20.2 20.9

6.6 2.9 6.3

23.0 5.0 8.5

43.8 21.1 18.0

ply becomes contaminated, the illnesses that result from the contamination will be clustered in people living within the distribution area of the water. Spot-maps on which cases are marked may show local concentrations that suggest possible sources. In interpreting such maps, it is important to relate the spatial distribution of cases to the density of population. The classical study of the 1854 cholera outbreak in the Golden Square area of London by John Snow used such a technique and led him to identify the particular water pump that was the source of the infection. In this instance, cases were clustered in the streets close to the Board Street pump, while comparatively few cases occurred in the vicinity of other pumps in the area. A special kind of locality difference is that which exists between urban and rural environments. In general, people who live in urban areas are subjected to different hazards from those experienced by people who live in rural areas. These differences alter their risk of certain diseases, sometimes to the advantage of the country person and sometimes to the benet of the town dweller. In urban areas, there may be better housing and sanitation but more overcrowding and air pollution; more leisure but less exercise, fresh food and sunlight; more industrial hazards but fewer risks of infection from animal contacts and vectors. In industrial societies, however, where commuting is a common practice, the distinction between town and country 20

dwellers is often blurred. Table 3.3 shows some differences in mortality between urban and rural areas in England and Wales.

Variations within single institutions

In institutions such as schools, military barracks, holiday camps and hospitals, variations in attack rates by class, platoon, chalet or ward may focus attention on possible sources or routes of spread. For example, in an outbreak of surgical wound infection, identifying the bed positions of patients, ward duties of staff and theatres used may suggest the identity of a carrier or other source of infection. Similarly, in places of work the danger of developing disease may be shown to be inversely related to distance from source of a chemical hazard. A high incidence of a disease amongst people who share the same environment does not prove that a factor within the environment was the cause of the disease. It may be that the people have chosen, or have been chosen, to share the same environment because they have an increased susceptibility to that disease or because of preexisting disease or disability.

Personal characteristics
The chances of an individual developing a disease may be affected by personal characteristics. The

Descriptive studies Chapter 3

Table 3.3 Differences in mortality amongst males between urban and rural districts in England and Wales 196973 (SMRs). Urban with populations Conurbations Malignant neoplasms Bronchus, trachea and lung Bladder Chronic rheumatic heart disease Ischaemic heart disease Inuenza Bronchitis Motor vehicle accidents Accidental poisoning Homicide Over 100 000 50 000100 000 Under 50 000 Rural

118 112 114 99 84 117 87 126 151

109 109 110 106 98 109 95 110 99

98 99 88 107 90 98 98 100 95

90 96 94 101 116 96 99 89 71

79 82 85 95 111 76 124 67 56

analysis of data on the incidence of disease in relation to the personal characteristics of victims provides useful indicators of possible causes. The personal characteristics can be classied as shown below.

Variation of disease due to personal characteristics

Intrinsic factors (affect susceptibility if exposed to causal agents) Age Gender Ethnic group Personal habits or lifestyle (affect exposure) Family Occupation and socioeconomic group

Intrinsic factors
Age Most diseases vary in both frequency and severity with age. In general, children are more susceptible to infectious diseases, young adults are more accident prone and older adults tend to suffer the results of long exposure to occupational and other environmental hazards. In infancy, immaturity and genetic defects affect susceptibility to disease. In later life, physiological changes, degenerative processes and an increased liability to malignant

tumours are the dominant determinants of the patterns of illness. The fact that the incidence of most diseases varies with age can complicate the comparison of morbidity and mortality between populations with dissimilar age structures. For example, the age structure of a population of military personnel is likely to be substantially different from that of a group of practising physicians. Therefore, it is to be expected that the two groups will differ in their incidence of many diseases. In order to make a valid comparison between these populations it is essential to adjust the data to take account of differences in their age structure. This procedure is called standardization (see Chapter 9). Age differences in the incidence of disease may also be accounted for by a so-called cohort effect. This occurs when individuals born in a particular year, or living at a particular point in time, are exposed to the same noxious agent. They then carry an enhanced risk of the disease caused by that noxious agent for a long period, sometimes for the rest of their lives. For example, the children who were exposed to radiation in Hiroshima and Nagasaki in 1945 when the atomic bombs were detonated have had higher than expected incidence of leukaemia throughout their lives.

Gender There is evidence that males are intrinsically more

21

Chapter 3 Descriptive studies

Table 3.4 Death rates at different ages for males and females in England and Wales, 2003 (deaths per 1000). Age Stillbirths Under 1 year 14 years 59 years 1014 years 1519 years 2024 years 2534 years 3544 years 4554 years 5564 years 6574 years 7584 years 85 and over Males 5.7 5.9 0.25 0.12 0.16 0.49 0.78 0.94 1.58 3.85 9.7 27.2 73.6 188.1 Females 4.9 4.5 0.20 0.10 0.11 0.24 0.27 0.44 0.94 2.54 6.0 17.0 50.5 159.8

these ethnic characteristics from a number of other factors which affect the incidence of disease, for example dietary habits, religious practices, occupation and socioeconomic status. The effect of ethnicity on the incidence of disease is best studied in communities where people of different groups are living side by side and in similar circumstances. For example, studies in the UK have shown a higher prevalence of type 2 diabetes in Asians compared with the white population. This is probably due to genetic differences. On the other hand, in New Zealand the differences in the cot death rate between Maoris and Europeans is related principally to the lower socioeconomic status of most Maoris and lifestyle factors such as maternal smoking.

Personal habits or lifestyle

vulnerable to disease and death than are females. This is rst apparent in the differential rates of stillbirth and early neonatal mortality, and remains throughout life (Table 3.4). Indeed, during later life, with the exception of disorders that are specific to the female, there are few diseases which have a greater incidence in women than in men. In most societies, men are exposed to a greater number and variety of hazards than are females often because of differences in their leisure and work activities. Even when the two sexes are exposed to the same hazards for the same period of time, there is evidence that women are less likely to develop disease and that they survive better than men. Some diseases appear to vary in incidence between the sexes only because they are more readily diagnosed in one sex than the other, for example gonorrhoea in men, or because they are more likely to come to medical attention, for example in mothers of young children.

Family Some diseases are especially frequent in certain families because of a common genetic inheritance, which is an intrinsic characteristic of the individuals. The risk of disease among members of the same family may also be increased because the members share a common environment and culture. Culture affects a wide range of disease-related factors such as type of housing, dietary habits and the way in which food is prepared, as well as the individuals reaction to illness. Occupation and socioeconomic group Some people are exposed to special risks in the course of their occupation. These include exposures to dust (particularly coal dust, silica and asbestos), toxic substances and gases used in industrial processes, and the risks of accident. Some occupations inuence habits such as the amount of tobacco smoked and of alcohol consumed or the regularity of meals, which in turn affect disease incidence. When interpreting any observed correlation between occupation and disease it is necessary to take account of the factors which determine a persons choice of occupation. Some may affect the persons susceptibility to disease; for example, tall and powerful people may choose physically demanding

Ethnic group This term tends to be used very loosely to describe a number of personal characteristics, including some that are strictly genetically determined, for example skin colour, and some that have nothing to do with genetics, for example country of birth and religion. It is often difcult to disentangle
22

Descriptive studies Chapter 3

Table 3.5 SMRs for ages 1564 years (England and Wales) showing trends by social class for specic causes of death. Cause of death (ICD number) Malignant neoplasm of stomach (151) Malignant neoplasm of trachea, bronchus and lung (162) IHD (410414) Cerebrovascular disease (430438) Bronchitis, emphysema and asthma (490493) I 50 53 88 80 36 II 66 68 91 86 51 IIIN 79 84 114 98 82 IIIM 118 118 107 106 113 IV 125 123 108 111 128 V 147 143 111 136 188

occupations whilst others may chose sheltered occupations because they already suffer mentally or physically disabling diseases. Some, because of chronic disease, may be unable to keep demanding jobs in the higher socioeconomic groups; they tend to move down the social scale (social class migration). Social class is derived from occupation and status within an occupational group (i.e. manager, foreman, unskilled). The concept of social class encompasses income group, education and social status, as well as occupation. Most diseases show a positive social class gradient, with a higher incidence in manual workers than in professional groups (Table 3.5).

Interactions of time, place and personal characteristics

Frequently, two or more factors correlate with the incidence of a disease and also with each other. It may be that only one factor is a causal agent or determinant and that the correlation with a second factor is fortuitous. Sometimes, however, two separate causes of disease interact with each other in such a way that the effect of the two acting together in the same individuals is greater than that of either acting alone. For example, while people who work with asbestos and who do not smoke have a higher incidence of bronchial carcinoma than other non-smokers, those who smoke have a much higher incidence than would be expected in people

with similar smoking habits in the general population. Interactions such as this are often very complex and the analysis of observed distributions can do no more than indicate possible determinants which merit more detailed and carefully controlled enquiry. Time, place and personal interactions can be separated if circumstances arise in which one of the variables can be kept constant while the others change. For example, comparison of disease frequency in migrant populations with the frequency in their place of origin is often informative, particularly where migrants move from an area with a high incidence of disease to one with a low incidence, or vice versa. When they migrate, they take with them their original hereditary susceptibilities but they change their risk of exposure to harmful agents. For example, the incidence of cancer of the stomach is higher in Japanese living in Japan than those living in the USA, while for cancer of the large bowel the reverse is true. In time, when migrants are assimilated into the host culture, they may be exposed to new risks in that culture. Thus, studies of migrant groups can also be used to measure the latent period between exposure and onset of disease. For example, the incidence of multiple sclerosis is higher in Europeans who migrated to South Africa before the age of 15 than in those born in South Africa. It must be stressed that caution is needed in studies of migrants because they are self-selected from the original population and their risks of disease may have been different from those who did not migrate.

23

Chapter 4 Surveys, survey methods and bias

Introduction
Many descriptive studies make use of routinely collected data. However, such data are often unsatisfactory for this purpose and specically designed surveys are needed. The problems are shown below.

managed or reported and therefore do not come to the attention of the medical profession. Example 1 Osteoarthritis is neither fatal nor is it always treated or reported. Studies of osteoarthritis based entirely on the cases treated in hospital or brought to the attention of the general practitioner may be misleading. Example 2 Acne varies considerably in its severity. In some individuals it is manifest by a few spots on the face or back whereas in the most severe cases there are widespread areas of pustules. Mildly affected individuals may ignore the manifestations, or use cosmetics or preparations available from pharmacies without prescription. The cases seen by the general practitioner will tend to be the more severe. However, some people will be more concerned than others and thus some mild cases may be seen by the GP. Specialist dermatologists will see the most severe and those refractory to treatment.

Problems with routinely collected data

Problems
Difculties in ascertainment of cases Variations in diagnostic criteria Absence of records of the attributes of individuals Unsuitable format of records Inconsistency in data presentation

Difculties in ascertainment of cases

The recorded number of patients with a condition may vary for reasons that have nothing to do with the actual frequency of the disease. For example, the tendency to seek medical attention and the availability of services may vary. This source of bias is of greatest importance when studying illnesses that are rarely fatal and therefore do not appear on death certicates, or that are not always medically 24

Variations in diagnostic criteria

These tend to vary between doctors and may change with time. This may be simply a matter of fashion or because the facilities for accurate diagnosis vary. Sometimes, there may be internationally agreed changes in classication practices. Example The ICD is revised about every 10 years

Surveys, survey methods and bias Chapter 4 and some diagnostic categories may not be carried forward from one revision to the next. In addition the diagnosis may involve a measurement that is not made routinely and/or recorded for the whole of the population. Example It is extremely difcult to study the epidemiology of hypertension in the community without doing special surveys because the denition of hypertension differs from one GP to another. By contrast, birth weight can be studied in some detail because all newly born babies are weighed and their weight is usually recorded. where these points are close together, it does not matter very much which is chosen, but in the case of chronic diseases the intervals may be months or even years. In such circumstances the reference point must be stated and be consistent. The above difculties with routinely available data can be partly overcome by well-designed routine information systems. Nevertheless, these cannot meet all requirements and many of the problems can only be overcome by surveys in which the data and means of collection are specied in advance and in which the study population is clearly dened.

Absence of records of attributes of individuals

The attributes of the individuals which the study proposes to investigate in relation to the presence of disease may not be recorded systematically. Example The occupation of patients is often not recorded or not recorded in sufcient detail in hospital notes to allow investigation of a cancer which it is suspected may result from occupational exposure to a carcinogenic agent.

Cross-sectional (prevalence) surveys

A cross-sectional (prevalence) survey is simply a descriptive study which, instead of relying on routine sources of data, uses data collected in a planned way from a dened population. The aim is to describe individuals in the population at a particular point in time in terms of their personal attributes and their history of exposure to suspected causal agents. These data are then examined in relation to the presence or absence of the disease under investigation or its severity with a view to developing or testing hypotheses as described in Chapter 3. Example A cross-sectional survey was carried out among a multiracial workforce at worksites in New Zealand by Scragg and colleagues between 1988 and 1990. The survey studied 5677 staff aged 4064 years. The subjects were asked about their age, ethnicity, past medical history, occupation and income. Their height, weight and blood pressure were recorded and an oral glucose tolerance test to detect diabetes mellitus was performed. The study showed that the prevalence of diabetes increased with age, was more common in Maoris and that approximately 50% of workers with diabetes were previously undiagnosed (see Fig. 4.1). The prevalence of diabetes was also signicantly correlated with weight and low income.

Unsuitable format of records

The data are recorded but are not usable because the form of the record is unsuitable, or because they are governed by strict rules of condentiality. Example Diagnoses may be recorded but not in a form or in sufcient detail to allow classication by ICD or other standard criteria.

Inconsistency in data presentation

In the analysis of deaths, the numbers and the date of occurrence are indisputable in countries where death registration is standard practice. However, when analysing morbidity by time, there are several possible points of reference. Those commonly used are the date of onset of the disease, the date of onset of symptoms, the date of rst diagnosis or the date of hospital admission. In acute diseases,

25

Chapter 4 Surveys, survey methods and bias

12 10 8 6 4 2 0 4044 4549 Age New European New Maori Known European Known Maori Figure 4.1 The prevalence of diabetes both known and previously undiagnosed in Maori and European workers. (From Scragg R et al. N Z Med J 1991; 104: 3957.) 5054 55+

Per cent with diabetes

Survey methods
A number of practical and theoretical problems can arise in the design and conduct of crosssectional surveys and other studies which may invalidate the results unless they are handled properly. The investigator needs to be aware of these potential problems and familiar with the methods that are available to solve them or to minimize their effects.

lecting a sample. Some commonly used sampling techniques are detailed below.

Types of sample
Simple random sample Systematic sample Stratied sample Cluster sample Multistage sample

Sampling
It is usually unnecessary to study the whole of a population in order to obtain useful and valid information about that population. The investigation of a sample has many practical advantages. In particular it reduces the number of individuals who have to be interviewed, examined or investigated. It is also often easier to obtain high response rates and high-quality information on smaller numbers. This is always preferable to poor-quality data on larger numbers. If a sample is used, it is essential to ensure that the individuals included in the sample are genuinely representative of the population being investigated the parent population. There are many methods available for se26

Simple random sample

In this sample each individual in the parent population has an equal chance (probability) of being selected. One way of obtaining a random sample is to give each individual a number and then to use a computer-generated table of random numbers to decide which individuals should be included.

Systematic sampling
This form of sampling is more convenient and is adequate for most purposes. People are selected at regular intervals from a list of the total population.

Surveys, survey methods and bias Chapter 4 It has the advantage of being easy for eld workers to use. Example If a 1 in 10 sample of school children is required then every 10th child on the school roll could be included. In some circumstances this method can lead to bias, for example when the school roll (or similar list) is compiled by class (or other grouping), which may affect randomness. mly selected from all possible groups of the same type, for example a random sample of all households in England as in the General Household Survey undertaken routinely by ONS. All members of the selected groups are included in the study. The underlying assumption is that the individuals belonging to any particular group do so for reasons unconnected with the disease being studied and the presence of any factor under investigation. The main advantage of this method of sampling is that the eld work is concentrated and therefore simpler and cheaper. The principal disadvantage is that diseases and associated factors themselves may have determined the group to which individuals belong, which the investigator may not suspect.

Stratied sample
In this sample the probability of an individual being included varies according to a known and predetermined characteristic. The aim of this method is to ensure that small subgroups which are of particular interest to the investigator are adequately represented. Example Suppose one of the attributes being investigated in a cross-sectional study of school children is the consequences of being an immigrant to the country. If immigrants comprise only 5% of the population, then a simple random sample would produce a group in which 5% are immigrants. Unless the sample is very large, the number of immigrants in the group may be insufcient for a conclusive analysis. To avoid this problem, the sample has to be weighted in favour of the selection of immigrant children. This is done by drawing separate random samples from amongst immigrant and indigenous children, e.g. 50% of immigrants and 10% of the indigenous group. Thus, all immigrant and all indigenous children have equal chances of selection although the chance of an immigrant being selected is greater than the chance of a locally born child being selected. When the data are analysed, the fact that the sample was recruited in this way must be taken into account.

Multistage sampling
This combines the above sampling techniques. For example, a series of clusters, say schools, might be identied and a random sample of them selected. Then within each school, a random sample of pupils stratied by class would be recruited to the study.

Bias in sampling
There are ve important potential sources of bias in selecting any sample. 1 Any deviation from the rules of selection can destroy the randomness of the sample. One of the most common temptations is to recruit volunteers to the study. This is in effect self-selection of participants and such individuals tend to be unrepresentative of the parent population. 2 Bias is introduced if people who are hard to identify in the parent population under study are omitted from the study. Thus, in investigating the health of school children the omission of children who are persistent absentees may seriously bias results if the reason for their absence is chronic illness. 3 The replacement of previously selected individuals by others can easily introduce bias. If, for example, it proves difcult to trace a person who has been selected or if that person refuses to cooperate, 27

Cluster sample
This involves the use of groups as the sampling unit rather than individuals (e.g. households, school classes or residents within blocks on a grid map). The groups to be studied should be rando-

Chapter 4 Surveys, survey methods and bias it is not acceptable to replace him or her with an easily traceable or cooperative individual. Replacement of a selected individual is acceptable only if it transpires that they were included in the sample in error, for example a selected subject was subsequently found not to satisfy study criteria. 4 If large numbers of individuals in the sample refuse to cooperate in a study, the results may be meaningless. Therefore, it is essential to make intensive efforts to enlist the cooperation of and trace all the individuals who have been sampled. 5 If the list of people used as a sampling frame is out of date, bias will be introduced owing to omission of recent additions and the inclusion of people who have departed. ments (questionnaires, laboratory or other measuring equipment) used. Without clarity of denition in the design of the study and consistency in its execution, errors will occur (see below).

Error and bias in numerator data

Subject variation Observer variation Limitations of the technical methods used

Subject variation
Differences in observations made on the same subject on different occasions may be due to many factors, including those outlined below. Physiological changes in the parameter observed, for example blood pressure, blood glucose. Factors affecting the response to a question, for example recollection of past events, motivation to respond and mood at time of interview, reaction to environment and rapport with interviewer. Induced changes because the subject is aware that he or she is being studied. (This is sometimes referred to as the Hawthorne effect.)

Error in rates
The analysis of epidemiological survey data usually entails the calculation of rates, for example incidence and prevalence rates, in exposed and non-exposed population groups. Rates may be affected by errors and bias in either the numerator or the denominator or both. Such errors can invalidate comparisons between rates, and result in misleading conclusions.

Error and bias in numerator data

The quality of numerator data is crucial for accurate classication of individuals according to their personal attributes, their exposure to suspected causal agents and whether or not they have the disease under investigation. In contrast to descriptive studies based on routine data, special surveys offer the investigator the advantage of being able to specify the observations that he or she wishes to be made, rather than being constrained by the data that are collected for other purposes. Furthermore, the investigator can prescribe the methods to be used in examining or questioning the individuals involved in the study. However, the investigator usually only has a single opportunity to make the observations on each subject. It is therefore essential that the information required is clearly dened at the outset and that efforts are made to ensure that consistent results are obtained by the instru28

Observer variation
The principal types of observer variation are as follows. Failure of the same observer to record the same result on repeated examination of the same material (inconsistency) this is called intraobserver variation. Failure of different observers to record the same result this is called interobserver variation. The greater the number of different observers, the greater are the chances of variation between them. Either of the above types of error can arise for several reasons. Bias induced by awareness of the hypothesis under investigation; for example, in a study of HIV infection, the observer may probe answers to certain questions more deeply if the subject has declared himself to be a homosexual or an intravenous drug user.

Surveys, survey methods and bias Chapter 4 Errors in executing the test or variations in the phrasing of a question, for example failure to be consistent in the use of a procedure, carelessness in setting up instruments or reading a scale, failure to follow instructions when administering a questionnaire, omission of some questions or tests, errors in recording of results. Lack of experience or skill, and idiosyncrasies of observers, especially when classication depends on a subjective assessment, for example misinterpretation by the interviewer of an answer to a question, lack of skill in the manipulation of instruments, poor motivation, lack of interest in the project. Bias in the execution of the test, for example preconception of what is normal or to be expected, digit preference (i.e. tendency to round off readings to whole numbers, 5s and 10s), inection of voice in asking questions. bias might arise. Some of the more straightforward principles are given below. The criteria used in diagnostic classication must be clearly dened and rigidly adhered to (even at the risk of missing a few cases). The features that must be present (or absent) for a diagnosis to be made must be specied. Classication of severity or grade of disease should be in quantitative terms where possible and it should cover the full range of possible types of case. All subjects should be observed under similar biological conditions on each occasion. Avoid uncomfortable circumstances. Design simple questions and use check questions for consistency of response. The number of observers used should be kept to a minimum. They should be trained properly to enhance their skills and test their variation on dummy subjects (or specimens). Take duplicate readings and record the mean value. Arrange for the classication to be reassessed by different observers, for example independent assessment of histopathology specimens by more than one pathologist. Where possible, subjects and observers should be unaware of (blinded to) the specic hypothesis under investigation in order that they are not inuenced by personal perceptions of the signicance of the variables being recorded. The tests selected should be relevant to the purpose. Those that give the most consistent results and are least disturbing to the subject are preferable. Equipment should be simple, reliable and easy to use. Test methods should be standardized by, for example, the use of standard reagents, sets of graded X-rays or slides, standard wording of questions and instructions on probing and interpretation of answers, and calibration of instruments against a standard reference. Quality control should be maintained to avert drift from standards.

Limitations of the technical methods used

Technical methods may give incorrect or misleading results for the following reasons. The test does not measure what it is intended to measure; for example, the presence of albuminuria in pregnancy, for which there are many causes, is a poor index on its own of the presence of toxaemia. Therefore, a study of toxaemia in which cases are identied solely by albuminuria will give misleading results. The method used is intrinsically unreliable or inaccurate, and thus yields results that are not repeatable or correspond poorly with those obtained by alternative methods, or do not correlate well with the severity of the condition being measured, for example peak ow rate in asthma. Faults in the test system, for example defective instruments, erroneous calibration, poor reagents, etc.

Avoidance of numerator error and bias

There are no hard and fast rules that can be applied to ensure that errors do not arise in surveys and that bias is avoided. Each project will require careful thought and consideration of where errors and

Error and bias in denominators

Errors occur when the population being investigated is not fully dened. Such errors can be mini29

Chapter 4 Surveys, survey methods and bias mized by making every effort to encourage cooperation of the potential subjects and avoiding any inconvenience or discomfort to them. All available means should be used to trace and persuade nonattenders to take part or continue to participate in the investigation. The similarity of those who participate and those who are lost from the study should be checked by comparing their general attributes such as age, marital status, sex and occupation to establish how representative they are of the total sample. There are several ways of handling people lost to follow-up in the analysis phase of an investigation. Exclude them from both the numerator and the denominator. Include them up to the time that they left. This involves calculating units of time at risk (see p. 33). Include all those lost for half the time at risk on the assumption that the rate of loss was even throughout the period and on average each individual was present for half the time. Analyse the data on the assumption that all those lost developed the disease, or had the most adverse outcome, and then on the assumption that none of them developed it. This will show the range within which the true result might lie.

Systematic error (bias)

This is a consistent difference between the recorded value and the true value in a series of observations. For example, if the height of an individual is always measured when the person is wearing shoes, then the measurement will be consistent but will have a systematic bias.

Discrimination
This relates to whether a test is able either to separate people with a disease (or a particular attribute) from those without the disease (or attribute) or to place subjects accurately on a range of severity (or a scale measuring an attribute). The degree to which this is achieved correctly is a measure of discrimination. A test with good discriminatory power has a small range of error in relation to the potential range of true results. There are two basic characteristics of a test which measure its discriminatory powers: its reproducibility and its validity.

Reproducibility (reliability, repeatability)

This is a measure of the consistency with which a question or a test will produce the same result on the same subject under similar conditions on successive occasions. A highly reproducible test must have low random error, although it may still have systematic error. When reproducibility is evaluated by retesting subjects, it is usually dened as the ratio of the number of cases positive on both occasions to the number positive on at least one occasion. It can be assessed by the following procedures. Replication of tests. The results of a series of measurements by the same observer or by different observers using the same test on the same group of subjects (or set of specimens) under identical conditions are compared. Comparison of test systems. The measurements are repeated using a different instrument or test system. Statistical analyses can be used to identify whether the variation is attributable to the test system, intraobserver variation, interobserver varia-

Assessment of error in surveys

Some terms that are frequently used in the assessment of error in surveys are given below. Common terms in survey error
Random error Systematic error Discrimination Reproducibility, assessed by: replication of tests comparison of test systems use of check questions random allocation of subject to interviewer Validity

Random error
This is due to the chance uctuation of recorded values around the true value of an observation. 30

Surveys, survey methods and bias Chapter 4 tion or subject variation. Similar methods can be applied to the assessment of the reproducibility of questions, but there are problems because when the same question is repeated, the subject (and observer) may be conditioned by replies given on previous occasions. Other procedures for assessing the reproducibility of questionnaires are: the use of check questions, i.e. questions which seek the same information though in a different form, for example age and date of birth; and the random allocation of subjects to different interviewers and comparison of results between groups.

Validity (accuracy)
This is a measure of the capacity of a test to give the true result. A valid test is one that correctly detects the presence or absence of a condition or places a subject correctly on a scale of measurement. For example, glycosuria as a test of the presence of diabetes has poor validity compared with a glucose tolerance test. Validity has two components. In the case of a test which divides a population into two groups, validity is assessed by how well it picks up those with diseases (its sensitivity) and how well it rejects those without disease (its specicity) (see p. 137).

31

Chapter 5 Cohort studies

Introduction
Cohort studies involve the investigation of groups of people who have no manifestations of the disease under study at the time they are recruited. The selected study group is observed over a period of time in order to measure the frequency of occurrence of the disease amongst people exposed to the suspected causal agent compared with its frequency amongst individuals who are not exposed. Cohort studies can be used in a similar way to identify the determinants of disease in the study population.

Types of cohort
Groups with special personal characteristics Groups with special exposures

constantly maintained (e.g. doctors, nurses, etc.); individuals who are members of a particular insurance scheme; or employees in industries with a low turnover of workers. At the time of recruitment to the study, the investigator identies the characteristics of the subjects by the use of standard questionnaires or the measurement of any number of biological variables. They are then followed up until a sufcient proportion have reached a predened end point (usually the development of the disease being investigated or death). During the follow-up period, their exposures to suspected harmful agents are recorded. Such cohorts can be used to estimate prevalence, incidence and risk in relation to a suspected causal agent without recruiting an additional comparison (control) group because the comparison group (those not exposed to the agent being investigated) is a subgroup of the cohort itself. This is called an internal control group.

Groups with special personal characteristics

Groups of individuals who have special characteristics unrelated either to their risk of exposure or to their risk of disease, and which make them easy to follow up, provide useful cohorts for the investigation of many diseases. Cohorts that have been used because they are easy to study over long periods of time include, for example, those selected because they belong to a profession of which a register is 32

Groups with special exposures

The other main type of cohort comprises groups of individuals who have all been exposed to the agent or the experience that is being investigated. This type of cohort requires the concurrent recruitment and study of an external control group. The control group in this instance must be drawn from a population that is similar to the exposed group in all respects other than their exposure to the agent

Cohort studies Chapter 5 under investigation. The data on the control group must be the same and collected in the same way as those on the exposed group. These two types of cohort study can be equally valuable in epidemiological studies. The choice depends on the question being studied and the availability of suitable study populations. exposed for over 7 weeks clearly have a greatly reduced risk of developing the dyscrasia since they have passed through the critical exposure period. If, in a study designed to assess the risk of agranulocytosis in patients exposed to chlorpromazine, the total number of treatment weeks is used as a denominator, it will give a distorted indication of the level of risk. In this case, the denition of exposure must specify the time period during which the individual consumed the drug.

Time at risk
In an ideal situation, all members of either type of cohort are recruited to a study at about the same time and followed up for the same period of time. Sometimes, it is not possible to recruit sufcient numbers to yield signicant results in a short period, particularly if exposure to the agent or the disease being investigated is relatively rare. Moreover, in most studies, some patients are lost to follow-up. Either situation will result in variations in the length of time during which individual members of cohorts are observed. This gives rise to problems in the analysis of the data. One way to handle variations in the periods during which individuals have been observed is to use the total time at risk in each group as the denominator. It is calculated by summing the units of time during which each person in the group was observed. It is expressed as the number of units of time at risk, for example 1 person-year = one individual at risk (or observed) for 1 year (or two people for half a year each). Caution must be exercised in the use of time at risk as a denominator. It is only valid if the risk of developing the disease in an individual is not inuenced by the period of exposure or the time at which the exposure occurred. If there is reason to believe that the risk of a disease is affected by the length of time an individual is exposed to an agent, the summation of the exposed time within a group will be misleading. For example, in 1969 Pisciotta demonstrated that chlorpromazine can cause agranulocytosis in some people and that it usually occurs after 57 weeks of continuous exposure (Pisciotta AV. JAMA 1969; 208: 1862). It follows that patients who are exposed for less than 5 weeks do not have the same risk as those exposed for longer, even though they might be susceptible. Patients

Advantages and disadvantages of cohort studies

Advantages
The main advantage of the cohort study design is that it is possible to distinguish antecedent causes from concurrent associated factors in the aetiology of disease. In both types of cohort study, the incidence of disease in exposed and non-exposed groups can be determined, allowing the calculation of absolute, relative and attributable risks (see p. 10). It is possible to study several outcomes from exposure to the same hazard. Bias in controls is less of a problem than in casecontrol studies because the necessary comparison groups (exposed and non-exposed) are built into the study design from the start. Even so, it is important to bear in mind that the two groups may not be equally susceptible to the disease under study. Because the study is prospective, it is possible to standardize methods, thereby reducing error due to observer, subject and technical variation (see p. 28).

Disadvantages
It is not possible to be certain that supposed aetiological factors are in fact causal. This requires experiments of a kind referred to in Chapter 3, which are rarely possible in human populations. Even with common diseases, large populations are usually required to obtain signicant differ33

Chapter 5 Cohort studies ences in incidence in exposed and non-exposed groups. Also, if the incubation period of the disease is prolonged, the results of the study may be greatly delayed. These factors tend to make cohort studies very expensive in resources. One of the major difculties encountered in cohort studies is in the follow-up of all subjects. Migration and withdrawal of cooperation may bias the results. It is necessary therefore to build into the study design a system for obtaining basic information on the personal characteristics and outcome of those who cannot be followed up in detail for the full duration of the study. This allows comparisons to be made between subjects who are fully studied and those who are not. In this way, serious selective bias may be detected and can be allowed for in the analysis and interpretation of the results. Finally, even though standard methods and diagnostic criteria are adopted, these may change owing to drift over a prolonged follow-up period and results in later stages may not be comparable with those obtained earlier in the study. on changes in smoking habits and other data were sent to the male doctors in 1957, 1960, 1972, 1978 and 1990. The fact that all the individuals being studied were doctors on the Medical Register aided follow-up considerably. Deaths of doctors are notied to the Medical Register, for reasons quite unconnected with the study, which enabled the investigators to follow up a cohort many years after it was recruited with comparative ease. The rst stage of the analysis was to divide the doctors into those exposed to the suspected harmful agent (smokers) and those not exposed (non-smokers). The mortality of the two groups was then compared. The conclusions of the investigators have had far-reaching consequences. An association was found between smoking and seven different cancers, most notably lung cancer, as well as with chronic obstructive lung disease, vascular disease, peptic ulcer and several other fatal diseases. The excess mortality was almost twice as high in the second half of the study as in the rst half (Fig. 5.1). It now seems that about half of all regular smokers will eventually be killed by their habit. There was a pronounced correlation between the death rate from lung cancer and the number of cigarettes smoked (Fig. 5.2). The data also revealed that the risk of death from lung cancer fell substantially in those who gave up smoking, a benet which increased with time. This study yielded two observations that could not have been made from descriptive studies alone. Firstly, the sequence of events was clearly identied smoking was followed by lung cancer, and secondly, a doseresponse effect was demonstrated. Both of these ndings weigh heavily in favour of the causal hypothesis. However, it should be remembered that the investigation was stimulated by the results of descriptive studies which showed a correlation between mortality from lung cancer and sales of cigarettes in England and Wales. The problem with a cohort recruited in this way is that if it is used to study the effects of an agent or factor which is very rare, or if the disease is a rare consequence of exposure, then the size of the cohort has to be very large in order to yield sufcient

Examples of cohort studies

Mortality in relation to smoking: 40 years observations on male British doctors (Doll
R, Peto R, Wheatley K, Gray R, Sutherland I. Br Med J 1994; 309: 90111) The classic study of the effects of smoking amongst British doctors is a good example of a study based upon a cohort that was used because it was administratively easy to identify and follow up. It involved the use of an internal control group. In 1951, the research team sent a simple questionnaire to all of the 59 600 doctors whose names were on the Medical Register of the UK at the time. The questionnaire enquired about their past and current smoking habits. Over 34 000 (69%) of the male doctors and more than 6000 (60%) of the female doctors who were contacted completed the questionnaire. The responding doctors were divided according to their past and current smoking habits and their subsequent mortality was recorded. Further questionnaires to obtain information 34

Cohort studies Chapter 5

Percentage of excess deaths in cigarette smokers

200

197191

100

195171

Figure 5.1 Age-specic excess mortality in cigarette smokers in rst half of study (lower line) contrasted with that in second half (upper line). An excess of 100% represents doubled death rate. Bars indicate standard deviations. (From Doll et al. 1994.)

3544

4554

5564 Age

6574

7584

8594

100

80

Per cent alive

60

40 Key 114 Cigarettes/day 1524 Cigarettes/day 25 Cigarettes/day Never smoked regularly 40 55 70 Age 85 100

20 Figure 5.2 Overall survival after age 35 years among cigarette smokers and non-smokers: life-table estimates, based on specic death rates for the entire 40-year period. (From Doll et al. 1994.)

35

Chapter 5 Cohort studies numbers of cases to detect a signicant difference between the risks in the exposed group and the non-exposed group.
Table 5.1 Average incidence of conrmed leukaemia in residents of Hiroshima and Nagasaki (194758) by city of exposure and distance from epicentre. Incidence rate per 1 000 000 person-years at risk Distance from epicentre (m) 0000999 10001499 15001999 20009999

Survivors of the Hiroshima and Nagasaki atomic explosions (Brill AB, Masanobu RR,
Heyssel RM. Ann Intern Med 1962; 56: 590609) The second type of cohort, one which is dened by the fact that the individual members have all been exposed to the same experience or agent, has the closest similarity to the laboratory experiment. There are many instances of cohorts that have been dened in this way. For example, the survivors of the atomic bomb explosions in Hiroshima and Nagasaki comprise a unique group of people who were exposed to high levels of ionizing radiation for a short time. In this group of people there was little difculty in calculating the proportion who developed leukaemia after, say 10 years, i.e. the absolute risk of leukaemia. However, in order to establish whether the incidence of leukaemia in the cohort was more or less than in a group not so exposed the relative risk of leukaemia it was necessary to study a group of people who were similar to the exposed group in all respects except for their exposure to ionizing radiation. In one of the many studies of the survivors from Hiroshima and Nagasaki, the control group comprised individuals who were living in the same area but outside the radiation zone. The study showed that the incidence of conrmed leukaemia was between 50 and 100 times greater in the exposed population than in the controls. Further investigations showed a clear relationship between the distance from the epicentre of the explosion and leukaemia incidence rates (Table 5.1), demonstrating a doseresponse effect.

Hiroshima 1366 308 42 28

Nagasaki 563 530 68 37

Table 5.2 Relative risk of breast cancer in women subjected to regular uoroscopies at different ages (risk in general population = 1). Age at rst exposure (years) <15 1519 2024 2529 3034 3539 40+ Relative risk 2.1 3.8 1.7 1.6 1.2 0.8 0.9

1940s and 1950s while being treated for pulmonary tuberculosis before the dangers of X-rays were fully appreciated. It has been shown that young women in the group that were irradiated had a higher than expected incidence of breast cancer. In this study, the control group was all other women of the same age in the population, the great majority of whom, it may be assumed, were not exposed to radiation in this way (Table 5.2).

Social class differences in IHD in men

(Pocock SJ, Shaper AG, Cook DG, Phillips AN, Walker M. Lancet 1987; ii: 197201) During 197980, 7735 men aged 4059 years were randomly selected from the lists of people registered with general practitioners in 24 towns in England, Scotland and Wales and were asked to participate in a long-term study. Seventy-eight per cent of those who were approached agreed to

Regular uoroscopy and risk of breast cancer (Boice JD, Monson RR. J Natl Cancer Inst
1977; 59: 82332) A cohort that experienced a different type of ionizing radiation is exemplied by the group of people who had large numbers of uoroscopies in the 36

Cohort studies Chapter 5

Table 5.3 Attack rates of major IHD events during follow-up, before and after adjustment for social class differences in risk factors.

Social class category I II III non-manual III manual IV V

Number of IHD cases 21 56 27 169 36 11

Cases per 1000 per annum Adjusted 5.6 5.2 6.0 8.2 7.4 5.6 Unadjusted 7.4 6.0 6.0 7.7 7.3 5.0

} }

5.5

7.9

} }

6.0

7.5

cooperate. These men were asked to complete a questionnaire which included questions on occupation, smoking habits and indicators of heart disease. They were also examined by a research nurse. Ninety-nine per cent of the men were followed up for an average of 6 years. The data were analysed using a multiple logistic regression model in order to adjust simultaneously the incidence rates of major IHD events for smoking, systolic blood pressure, serum cholesterol, age and social class. Of these men, 336 experienced major IHD

events (dened as fatal IHD or myocardial infarction). The crude attack rates and the attack rates adjusted for the risk factors set out above are shown in Table 5.3. The results indicate that, after taking account of the differences in smoking habits, systolic blood pressure, serum cholesterol and age between the social class groups, there remains an unexplained higher incidence of major IHD events amongst men in manual occupations compared to those in non-manual occupations.

37

Chapter 6 Casecontrol studies

Introduction
The basic principle of a casecontrol study is to compare the personal characteristics and exposures of individuals with and without the disease (or outcome) of interest. The individuals with the disease are referred to as cases; those without are controls. Both the validity and generalizability of a casecontrol study are affected by the way in which cases and controls are identied and recruited.

Identication and selection of cases

The value of a casecontrol study is profoundly inuenced by the ways in which both the cases and the controls are selected. Ideally, all the cases of the disease in a dened population should be included in the investigation. However, it is rarely feasible or indeed necessary to do this in order to reach sensible and valid conclusions. Most studies are implicitly or explicitly concerned with a sample of cases. The source of potential cases is determined to a great extent by the disease that is being investigated. When admission to hospital for treatment of the disease is inevitable (e.g. breast cancer, fractured femur, intestinal obstruction, end-stage renal failure) then cases can be recruited from hospital inpatients. However, it is possible that although hospital admission for the treatment of a condi38

tion is inevitable there are some diseases that result in sudden death before the patient can be treated (e.g. acute myocardial infarction). In these circumstances it is necessary either to accept that the ndings can only be applied to those who survive long enough to be admitted or to devise a method of incorporating the fatalities into the study. For many diseases hospital admission is not inevitable. It used to be standard practice to admit everyone suspected of having a deep venous thrombosis (DVT). During the admission they would be investigated and those in whom the diagnosis was conrmed would be started on anticoagulant therapy before discharge. Nowadays it is not unusual to investigate the individual as an outpatient, initiate therapy on an outpatient basis and then hand the long-term management over to the GP. It follows that, although in the past it would have been legitimate to recruit cases of DVT from hospital inpatients, it is no longer the case. The investigator thus has to identify an unbiased source of cases, that is, a clinic or process through which all or most genuine cases will pass. In the case of DVT this could be anticoagulant clinics or general practice records of patients with a diagnosis and anticoagulant treatment. In some diseases the majority of cases are treated by the GP with or without referral of a few to a specialist. Lower urinary tract symptoms indicative of benign prostatic hyperplasia is an example of this situation. GPs manage many men with alpha-

Casecontrol studies Chapter 6 blocking agents or 5-alpha reductase inhibitors. Those who fail to respond may be referred to hospital for catheterization or surgery. For conditions such as this the general practice is often the most appropriate source of cases for an investigation. For most studies it is desirable to recruit incident cases (cases newly presenting with the disease) rather than prevalent cases (cases with established disease). The problem with prevalent cases is that they represent the survivors from an earlier cohort of incident cases. Thus, in a study of the factors that lead to end-stage renal failure, it would be inappropriate to recruit the cases for study from amongst patients undergoing regular dialysis at the time of the investigation. Patients who have had regular dialysis for, say, 1 year will be the survivors of a larger group who started dialysis a year before. The survivors are unlikely to be representative of all of those who started treatment and therefore the ndings from any study of them might have limited value. Even when incident cases only are recruited to a study the stage of the disease at which they present will vary. This should be taken account of during the analysis of the data. For example, women may rst present with breast cancer at any stage, including metastatic disease. People registered with the same general practice. Students at the same school or educational institute. Workers in the same factory or ofce. People on the same electoral or other population register. Patients being treated in the same hospital. People randomly selected from the community in which the case lives. This can be achieved by, for example, random digit dialling. Using this method, controls that fail to meet basic recruitment criteria are discarded after a few key questions and the remainder are included in the investigation. Hospital patients, either all attenders or, more usually, those with conditions believed to be unrelated to the factors under study. The main limitation of using hospital patients as controls in any study is that, even though they may not have the disease being investigated, they are unlikely to be a random sample of the general population from which the cases are drawn. For example, even though a hospital patient does not have the disease under investigation he or she may have another disease caused by the same agent or whose presence could have affected exposure to the causal agent of the disease under investigation. Moreover, people who live in poor social environments are more likely to be admitted to hospital than those who live in better circumstances and their use as controls may introduce a social class bias. Controls may be selected as a group or selected for each of the cases as they arise. Thus, it is legitimate to recruit, say, 50 cases from a particular community and the same number or more controls from the same community. It is also legitimate to recruit one or more controls per case as they arise without reference to any personal characteristic. In a matched casecontrol study the controls are recruited for each case on the basis that they share one or more characteristics. In many ways the matched casecontrol study is more efcient; however, it will not provide any information on the variable or variables used for matching. Thus, in an investigation of the effect of oral contraceptives on the risk of venous thrombosis it might be appropriate to match the controls to 39

Selection of controls
Control subjects are essential in order to establish the frequency with which the suspected causal agents or determinants occur in people who do not have the disease under investigation. Control subjects must not have the disease being investigated at the time the case arose (index day). Controls must be a representative sample of the population from which the cases were recruited and thus are at similar risk of having been exposed to the suspected agent. Once selected, controls should be neither discarded nor replaced for any reason other than that they fail to meet the selection criteria; for example, if they were mistakenly drawn from another population. Depending on the condition being investigated controls could be selected from amongst:

Chapter 6 Casecontrol studies the cases on age. When doing this the study will provide no information on the effect of age on risk. Likewise a study could be designed to identify the risk factors for acute myocardial infarction where the controls are matched to the cases by age, gender and body weight. Such a study would provide no information on the effect or age, gender or body weight on the risk of acute myocardial infarction. This is acceptable if the effects of the matched variables are well known and therefore of little immediate interest. It is important to note that error (or bias) in the selection of controls will have exactly the same effect on the outcome of the study as bias in the selection of cases. In casecontrol studies (and in cohort studies that involve selection of external controls) as much attention must be given to the identication of, and collection of data from, the control subjects as is given to the cases. causal agent and those not exposed. By contrast, casecontrol studies only provide data from which the rate of exposure to suspected harmful agents in diseased and non-diseased individuals can be calculated. This means that neither the absolute risk nor the attributable risk nor the precise relative risk resulting from exposure can be calculated. The difculty is shown schematically in Tables 6.1 and 6.2. In a cohort study, the subjects studied are all those exposed (A + C) and all those not exposed (B + D) to the suspected causal agent (Table 6.1). The subjects subsequently reveal themselves as diseased or non-diseased within these categories. There is therefore no difculty in calculating the disease rate in the total population (A + B/A + B + C + D) or in the exposed persons (A/(A + C)) and those not exposed (B/(B + D)). The subsequent calculation of relative risk (RR) and attributable risk (AR) presents no problem: RR = A ( B + D) A B divided by = A+C B + D A ( A + C) AR = A B minus A+C B+D

Collection of data from cases and controls

Data can be obtained by interview, questionnaire or reference to pre-existing records or a combination of one or more of these sources. Whatever the method of obtaining information on the personal characteristics and exposures of the cases and controls the same method must be used for both. Wherever possible the eld workers should be unaware of the status (case or control) of the subject about whom they are collecting the data.

The subjects in a casecontrol study are identied either because they have the disease (the cases) or because they do not have the disease (the controls) that is being investigated. They are subsequently subdivided into exposed and not exposed subgroups (Table 6.2). It is not possible to
Table 6.2 Division of subjects in a casecontrol study.

Risk in cohort and casecontrol studies

Cohort studies are designed to provide the data needed to calculate incidence rates of the disease amongst individuals exposed to the suspected
Suspected cause present Suspected cause absent Total

Diseased a b a+b

Not diseased c d c+d

Table 6.1 Information available in cohort studies. Disease present Exposed to suspected cause Not exposed to suspected cause Total A B A+B Disease absent C D C+D Total A+C B+D A+B+C+D

40

Casecontrol studies Chapter 6 derive the total numbers of cases in the population who were exposed and not exposed because a casecontrol study is not based upon a known proportion of the population in either category. Consequently, neither the incidence rate in the population as a whole nor the incidence rate amongst the people exposed to the suspected harmful agent can be derived. It follows from this that risk cannot be calculated directly. However, an approximation of the relative risk can be derived from casecontrol data. This approximation, although often referred to as the relative risk, is more correctly termed the odds ratio (OR). It is calculated as follows. Using the notation in Table 6.1 the true relative risk is: A ( B + D) A ( A + C) In the case of most diseases, the proportion of the population who are affected, whether or not they are exposed to the suspected causal agent, is small. Thus, A is small in relation to C; likewise B is small in relation to D. It follows that D will approximate to B + D and C will approximate to A + C. The approximation to the relative risk, the OR, then becomes: AD ad = BC bc This approximation to relative risk is used in all casecontrol studies but it is only valid if the incidence of the disease is low. In most circumstances it is not possible to calculate attributable risk from a casecontrol study. Thus the increase in risk of heart disease associated with alcohol may be the result of smoking. There are two ways to take account of confounding variables. 1 Analysis of subsets of the data (partitioning) dened by the confounding variable. This can be illustrated by considering a study of the effect of age at rst birth on womens risk of carcinoma of the breast. Women who have their rst child whilst young tend to have more children than women whose rst child is born late in their reproductive life. It follows that if there is a statistical association between age at rst birth and the risk of breast cancer it is likely that there will also be an association between family size and risk of breast cancer. The two effects can be separated by restricting the analysis to women who have had only one child (thereby separating out the effects of parity) and calculating the risk according to age at rst pregnancy. Alternatively the analysis can be restricted to women who had their rst child at a given age and the risks calculated according to parity. The disadvantage of this technique is that not all the available data can be used in the critical analyses. 2 Use of a multivariate analysis technique to adjust the relative risk for the effect of confounding variables. The most usual statistical model is logistic regression. An advantage of this method is that it allows simultaneous adjustment for the effects of more than one confounding variable. Similar methods are used to adjust for the effects of confounding variables in cohort studies.

Effects of high incidence of exposure

Essentially, the success of a casecontrol study is dependent upon there being a signicant difference between the proportion of cases exposed to the suspect agent and the number of controls so exposed. If the incidence of exposure is very high, it may be impossible to demonstrate such a difference. Consider an extreme example of a casecontrol study designed to identify the possible causal agents of carcinoma of the bronchus which is conducted in a population where the prevalence of cigarette smoking over the age of 15 years is 100%. In 41

Confounding variables
A confounding variable is a characteristic or exposure that is associated with both the exposure being investigated and the outcome (disease). For example, in an investigation of the association between alcohol consumption and the risk of heart disease it is likely that amongst the cases there would be a greater proportion of smokers amongst the cases than amongst the controls. This is because smoking is correlated with alcohol consumption and smoking is associated with heart disease.

Chapter 6 Casecontrol studies such a situation there can be no difference between the proportion of cases and controls exposed to cigarettes. It follows that smoking could not be revealed as a risk factor. It does not follow that no risk factors will be revealed by such a study, but those identied may be associated rather than causal and the principal cause will be missed. people who choose to smoke are also constitutionally predisposed to lung cancer or are exposed to another noxious agent more often than are non-smokers. This problem is less conspicuous when dealing with highly specic agents such as microorganisms or in situations where the time between exposure and onset of symptoms is short. There are sometimes difculties in selecting and recruiting appropriate controls. This is important because the value of the results obtained from a casecontrol study is as dependent upon the proper selection of controls as it is on the identication of affected individuals. Because casecontrol studies are not based on dened populations, the incidence of the disease within that population cannot be calculated from the study.

Advantages and disadvantages of casecontrol studies

Advantages
Despite the approximations that have to be made in the analysis of casecontrol studies, they do have some important advantages over cohort studies. By concentrating effort on the identication of affected individuals and recruiting controls from the unaffected population, the number of subjects required to obtain signicant results is kept to a minimum. Results can be obtained relatively quickly because the investigation does not have to wait for the disease to develop, as it does in cohort studies. This means that it is a relatively inexpensive type of study.

Examples of casecontrol studies

Sexual activity, contraceptive method, genital infections and cervical cancer
(Slattery M, Overall JC, Abbott TM et al. Am J Epidemiol 1989; 130: 24858) It has been suggested that cervical cancer is a sexually transmitted disease. Between 1984 and 1987 a casecontrol study was carried out in Utah, USA, where a high proportion of the population are active members of the Church of Jesus Christ of Latter Day Saints (Mormons). The study was designed to explore the relationship between cervical cancer and sexual activity, the use of barrier methods of contraception and certain types of genital infection. The subjects were women aged 2059 years, newly diagnosed with cervical cancer. Controls were identied by use of a random digit dialling telephone sampling technique. They were matched to cases by 5-year age intervals and county of residence. Interviews were completed for 266 women with histologically conrmed carcinoma in situ or invasive squamous cell cervical cancer and for 408 matched controls. After adjustment for age, education, church attendance and cigarette smoking, by means of mul-

Disadvantages
Casecontrol studies generally rely upon retrospective data, which have their own inherent problems. The ability of individuals to recall past events tends to be unreliable due to a tendency for memory to be selective. Records of past events may be incomplete in respect of variables that are the subject of investigation, and the ways in which the relevant observations and measurements were made are not usually standardized. This gives rise to uncertainty regarding their validity. Because the data are collected after the event (retrospectively) it is difcult to be sure whether a demonstrable correlation is causal or not. Thus, the nding that a history of cigarette smoking is common amongst individuals with lung cancer does not prove that the former preceded and caused the latter. Alternative explanations are that 42

Casecontrol studies Chapter 6 tiple logistic regression models, several signicant risk factors were identied. These included multiple sexual partners, current mate having multiple sexual partners, reported Trichomonas infection and serological evidence of herpesvirus type 2 infection (Table 6.3). It should be noted that there is a pronounced gradient of risk relating to increased numbers of partners and increased numbers of partners of the mate of the woman. A protective effect was noted from use of foam or jelly as a contraceptive method (OR = 0.44), and from the use of diaphragms (OR = 0.67) or condoms (OR = 0.53) in women who reported more than one sexual partner. These data lend support to the hypothesis that cervical cancer is due to a sexually transmitted agent. impractical and unethical to conduct a large-scale randomized control trial to test the validity of this suggestion. Therefore, a casecontrol study was set up which aimed to identify all children admitted to hospitals in the UK with serious acute neurological illnesses of the types which it was suggested could be caused by the vaccine and lead to permanent brain damage. For each case child reported, two control children, matched for age and sex, were selected from those living in the same local area. The past histories of immunization and of other possible predisposing or aetiological factors were obtained for both case and control children in identical manners. Of 904 cases of encephalopathy and severe convulsions reported, 30 (3.3%) had received pertussis vaccine within 7 days before becoming ill, compared with 23 (1.3%) of 1783 control children immunized within 7 days before a dened reference date, which was a signicant difference (OR = 3.3) (Table 6.4). The children were followed up a decade later to determine the late outcome of their illnesses. Nearly two-thirds either had died or were suffering from signicant neurological dysfunction. Of 367 such children, 12 cases (3.3%) were pertussis vaccine associated compared with 6 (0.8%) of 723 controls, which gives an OR of 5.5. Thus, the study showed that there is a small

Pertussis immunization and serious acute neurological illnesses in children (Miller DL,
Madge N, Diamond J, Wadsworth J, Ross EM. Br Med J 1993; 307: 11716) In 1975, widespread public alarm was created by the suggestion that whooping cough vaccine might cause severe encephalopathic illnesses followed by permanent brain damage in a small but signicant number of children. It would have been
Table 6.3 Risk factors for cervical cancer.

Numbers Risk factor Number of sexual partners of woman <2 23 45 610 > 10 Number of sexual partners of mate 1 23 45 610 > 10 Trichomonas infections Herpes type 2 (neutralization index > 1000) Cases Controls

Odds ratio Crude Adjusted

25 54 47 57 69

210 73 53 39 28

1.00 6.21 7.44 12.27 20.70

1.00 3.43 3.59 5.51 8.99

24 42 38 45 49 53 12

198 78 39 22 23 21 6

1.00 4.44 8.03 16.87 17.57 4.61 6.57

1.00 2.72 4.99 7.98 8.9 2.10 2.70

43

Chapter 6 Casecontrol studies

Table 6.4 Pertussis vaccine and encephalopathy in children. All outcomes n Cases Controls Odds ratio 904 1783 Vaccine associated (%) 30 (3.3) 23 (1.3) 3.3 Dead or dysfunctional 10 years later n 367 723 Vaccine associated (%) 12 (3.3) 6 (0.8) 5.5

but denite risk of serious acute neurological illnesses after whooping cough vaccine, though the risk was much smaller than some workers had suggested from totally uncontrolled series of cases. It was also clear that children who suffered from such illnesses often died or had signicant long-term sequelae, though the number of such cases associated with recent pertussis immunization was too small to be certain that the vaccine was on its own responsible for cases of permanent brain damage. This study illustrates the difculty of identifying aetiological factors in extremely rare conditions.

hazards appeared to be the leather, or adhesives used, or both.

Results of casecontrol study of leukaemia and lymphoma among young people near Sellaeld nuclear plant in West Cumbria
(Gardner MJ, Snee MP, Hall AJ, Powell CA, Downes S, Terrell JD. Br Med J 1990; 300: 4239) Concern about levels of childhood cancers around nuclear installations and a consequent public enquiry led to several studies being set up. One was a casecontrol study of leukaemia and lymphoma among young people living in the vicinity of the Sellaeld nuclear plant in West Cumbria. Its aims were to explore whether known causes or factors associated with the nuclear site were responsible for the apparent excess. A total of 74 cases of leukaemia and lymphoma among people born in West Cumbria and diagnosed there at ages under 25 between 1950 and 1985 was identied. Risk factors in cases were compared with those in up to eight controls matched by date of birth and sex, selected from the same birth register as their respective cases. The expected association with prenatal exposure to X-rays was found, but the main nding was of signicantly raised relative risks in children born near Sellaeld and in children whose fathers were employed at the plant (RR = 2.4), particularly those fathers with high radiation dose recordings before the childs conception (RR = 6.4). At the time, no other satisfactory explanation was put forward and it was concluded that ionizing radiation may be leukaemogenic to offspring. This interpretation has been subsequently challenged in the scientic literature.

Perinatal deaths and maternal occupation

(Clarke M, Mason ES. Br Med J 1985; 290: 12357) Reproductive hazards are thought to exist in many industries. In order to explore this problem, a casecontrol study of perinatal death occurring in Leicestershire was carried out between 1976 and 1982. Case notes were reviewed and the mothers were interviewed in all 1187 cases of perinatal death during this period. The control for each case was selected as the next live birth occurring at the place or intended place of delivery. All maternal and paternal occupations and industries were recorded at the interview with the mother. A total of 671 mothers were employed outside the home at some time during pregnancy. An analysis of maternal occupations showed that the OR for the risk of perinatal death was exceptionally high in women employed in the leather industry (OR = 2.1 after adjustment for social class). A similar excess was found in all towns within the county where shoe manufacture took place. No other risk factor was found to account for this observation. Possible

44

Chapter 7 Intervention studies

Introduction
Descriptive, cohort and casecontrol studies are used to develop and test hypotheses about the possible causes and determinants of disease. The results may suggest methods of prevention or treatment which then need to be formally evaluated. Intervention studies are most often used for this purpose and are commonly called clinical trials. They are essentially experimental studies in which the efcacy and safety of disease management are evaluated by comparing the outcome of the intervention in test and control groups. The intervention under test is most often a new preventive or treatment regime, but the method can also be used to compare different established regimes and to evaluate the effectiveness and efciency of different forms of service provision. Experimental studies in which the incidence of a disease in those deliberately exposed to a suspected causal agent, or protected from it, is compared with that in controls can also be of value and provide the most convincing evidence of a causal relationship. However, for practical and ethical reasons this approach is rarely adopted except in animal studies.

Methods in clinical trials

The methodologies of preventive and therapeutic trials have some similarities to those used in cohort studies. The basis of a clinical trial is the random al-

location of individuals in a population to test and control groups. The intervention under trial is applied to the test group but not to the control group and the effect is assessed in terms of the same dened outcome in both groups. The selection of the study population should be governed by the following considerations. The population under study should be representative of the population in which it is intended to apply the intervention being tested (this is called the target population). It is important to choose a stable population in which there are unlikely to be heavy losses during the follow-up period and whose cooperation is reasonably assured. Volunteers are usually not acceptable since they tend to differ from non-volunteers in important respects, such as motivation and past history of illness. The likely frequency of the outcome being measured should be known, since this critically affects the required sample size. The population should be readily accessible. Trials are often most conveniently conducted in patients attending general practice or hospital, or in residential institutions, factories, the armed forces, etc., even though they may not fairly represent the general population in all respects.

45

Chapter 7 Intervention studies

Allocation to test and control groups

In principle, the allocation of patients to test and control groups should be random: hence the term sometimes applied to clinical trials, randomized controlled trials (RCTs). The aim is to ensure that those treated and those untreated are similar in all respects prior to intervention. This is necessary to guard against the possibility that some factor other than the intervention could account for differences in outcome in the two groups.

Stratied allocation
Where the population is relatively small and nonhomogeneous, random selection within specied subgroups, for example age groups, may be desirable to ensure similarity in relevant characteristics between test and control groups.

No allocation
Sometimes random allocation of treatment would not be ethical, for example a trial of a new type of measles vaccine in children. In this case, the comparison must be with past experience or that in other populations. It is difcult in such cases to measure the extent of any benet with condence. Since willingness to cooperate may not be randomly distributed in the population, allocation should be deferred until agreement to participate has been obtained. To avoid bias in reporting illnesses and other possible behavioural differences, subjects should not know to which group they have been allocated. In drug and vaccine trials, this often entails the use of a placebo treatment for controls which must be presented in an identical form to the active treatment. In the case of some procedures, for example provision of different types of service, blind allocation is not possible. A trial in which neither the subject nor the people assessing outcome know whether the subject is receiving active treatment or not (or which of two different treatments is being given) is called a double-blind trial.

Types of random allocation

Individual Cluster: of whole groups or communities Stratied: random selection within specied subgroups

Individual allocation
The allocation of individuals to test or control groups must be random. Other methods of assignment to treatment groups are to be avoided. For example, alternate allocation to test and control may enable the patient or the person who assesses the outcome to guess the group to which the subject has been allocated.

Cluster allocation
For practical reasons, allocation is sometimes made of whole groups or communities. This is because, for example in trials of a vaccine, the spread of infection may be inhibited in unvaccinated people if a proportion of the population is protected, thereby obscuring the benet derived from vaccination. Similarly, in recent trials of preventive advice against coronary heart disease, the test and control groups were workers in randomly allocated factories, in order to minimize contamination of the control group with advice offered to the intervention group. 46

Outcome
The outcome to be assessed must be specied in advance. It should be expressed in terms of advantage to the patient or to the community, for example reduced incidence or severity of disease or cost to the health service. Assessment criteria should be clearly dened, consistently applied and reliably recorded in order to minimize bias in the measurement of outcome. Misclassied cases in either test or control group will reduce the size of difference between them in the incidence of disease and thus

Intervention studies Chapter 7 give a spuriously reduced apparent benet from the treatment. The safety of an intervention is as important as its efcacy and the assessment of outcome should always include the frequency of adverse effects of the intervention as well as its benets.

Sequential analysis
Sometimes, when a result is required urgently or when the anticipated benets are high or the possible adverse effects are serious, the results are analysed sequentially. This technique involves continuous data analysis and allows the trial to be stopped immediately when a signicantly benecial or adverse effect has been demonstrated. Trials can also be stopped when the results fail to reach a previously dened level of signicance, usually that which the investigators regard as the minimum useful benet.

Follow-up
Procedures for the follow-up of subjects in both test and control groups should be the same, giving particular attention to the following. The data collected must be obtained and recorded in a standard manner. The method used should be simple and should be sufciently sensitive to detect reliably all relevant events in members of the study population. Follow-up must be equally rigorous in both test and control groups. Follow-up must start from the time of allocation and continue for long enough to evaluate fully the outcome in all subjects. Cooperation must be maintained at the highest possible level, and losses from the study population for any reasons should be minimized.

Ethical considerations
The ethical questions that arise in the planning and conduct of RCTs are shown below.

Ethics
What are the possible risks of treatment and of failure to treat? Is it right to expose some people to possible harm from an untested treatment or to withhold from others a possibly benecial treatment? Is it right to introduce a new treatment into use without rst assessing its safety and benets by a properly conducted trial? To what extent should a trial be explained to the subjects before they agree to participate? How can the welfare and safety of participants be safeguarded while preserving the principle of blind assessment?

Analysis
It is usual to use an intention to treat (ITT) analysis with randomized controlled trials. This method requires that study subjects are analysed according to the treatment allocated irrespective of whether it was taken. The reason for this is that the treatment allocated is random and changes in treatment (including non-compliance) are not. Therefore, there is a potential for confounding if comparisons of the actual rather than assigned treatment are undertaken. For example, consider two treatments where one is superior to the other: in particular, it acts faster. If the analysis is carried out on patients who comply with the treatment plan then the superior treatment may appear to be worse simply because the successes early in the programme are not included in the analysis.

Before a new drug can be licensed there is a requirement to perform a series of RCTs and demonstrate its efcacy (or relative efcacy) and its safety. The trials must be carried out in accordance with protocols set out by the licensing authority. The general method of the RCT is increasingly being applied to interventions other than medicines. All studies involving humans must be approved by a recognized ethical committee.

47

Chapter 7 Intervention studies group to which the patients were allocated, but the patients did not. This was to enable the doctor to adjust drug dosage in those on active treatment if necessary to achieve the target level of blood pressure. All other management was the same in both treatment and placebo groups. Recruiting took place over 9 years and the data were analysed sequentially every 6 months in order to test whether any signicant differences were emerging. In the end, 85 572 person-years of observation accrued. There was a very signicant reduction in the incidence of stroke in the treated group, but no difference in the rates of coronary events (Table 7.1). The overall incidence of cardiovascular events was reduced, but mortality from these and all causes was not. It was concluded that if 850 mildly hypertensive patients are given treatment for a year, about one stroke will be prevented. On the other hand, this would subject a substantial percentage of patients to chronic side-effects, most but not all of which would be minor.

Examples of intervention studies

MRC trial of treatment of mild hypertension (Medical Research
Council Working Party. Br Med J 1985; 291: 97104) It has long been known that people with high blood pressure have an increased risk of stroke and other cardiovascular events and that treatment is effective in reducing the incidence of these conditions in severe hypertension. However, the value of treating mild hypertension compared with disadvantages of long-term therapy in otherwise healthy people was less certain. An RCT of treatment in such cases was therefore carried out by the MRC. Even though hypertension and cardiovascular complications are relatively common conditions, it was calculated that this would require a very large-scale trial in order to obtain a statistically signicant result. Subjects for the trial were found by screening blood pressure measurements in 515 000 people aged 3564 years selected from the agesex registers of 176 general practices in England, Scotland and Wales. In this way 17 354 patients with a diastolic pressure in the range 90109 mmHg and systolic pressure below 200 mmHg were identied. Patients were randomly allocated to one of four groups, two of which were treated with different hypotensive drugs and two with placebo tablets which looked identical to the active drug tablets. Randomization was stratied by age and sex. The target level of blood pressure was below 90 mmHg to be reached within 6 months of entry. The study was singleblind only, i.e. the doctor knew the treatment

Prevention of rickets in Asian children: assessment of the Glasgow campaign

(Dunnigan MG, Glekin BM, Henderson JB et al. Br Med J 1985; 291: 23942) There have been many reports of vitamin D deciency leading to rickets in infants and school children, and osteomalacia in women among the British Asian community. Theoretically, this would be easily remedied. The treatment is clear cut and no trial of the efcacy of vitamin D is needed. The acceptability of a prophylactic programme and its effectiveness in reducing the prevalence of rickets, however, needed to be assessed. Random alloca-

Table 7.1 Mild hypertension: main events in treatment and control groups. Active treatment Event difference Stroke Coronary events All cardiovascular events All cardiovascular deaths Non-cardiovascular deaths Number 60 222 286 134 114 Rate 1.4 5.4 6.7 3.1 2.7 Placebo Number 109 234 352 139 114 Rate 2.6 5.5 8.2 3.3 2.7 Percentage difference 45 6 19 4 0

48

Intervention studies Chapter 7 tion of individuals to treatment and control groups would be inappropriate and unethical and, in such circumstances, though less than ideal, a beforeand-after intervention assessment is often used. This study reported on the results of a campaign to promote the use of vitamin D supplementation in Glasgow. In a precampaign survey, blood samples were obtained from 189 children aged 517 years and those with biochemical evidence of rickets had an X-ray examination of the knees. In postcampaign surveys, 2 and 3 years later, 255 children were similarly examined. On both occasions the children were asked about their frequency of consumption of vitamin D supplements (in younger children this was checked with mothers). The results showed a striking reduction in the prevalence of rickets in children who took regular or even intermittent vitamin D supplements, and the number of hospital discharges of Asian children with rickets in Glasgow declined rapidly after the start of the campaign. Clearly, the decline in rickets could have been due to factors other than the ofcial vitamin D supplement campaign, for example increasing adoption of a Western diet and lifestyle. However, the time and place reduction in rickets prevalence, backed by objective measures, lends support to an assessment of the effectiveness of the campaign. condently advised to give up smoking in the expectation that their health and prognosis will improve. However, ex-smokers are not a random sample of former smokers and their reasons for giving up may be related to other factors which inuence their risk of developing smoking-related diseases. Nor is it certain how effective antismoking advice is in inuencing smoking behaviour. Therefore, in 1968 the authors set up an RCT of antismoking advice in 1445 male smokers, aged 4059 years, at high risk of developing cardiorespiratory disease. They were allocated at random to an intervention group who were given individual advice on the relationship of smoking to health and challenged to consider their situation. Those who declared a wish to stop smoking were given support and encouragement for an average of four further visits over 12 months. The control group were given no specic advice. All subjects completed a questionnaire 1, 3 and 9 years later. Deaths in the group were monitored. After 1 year, the reported cigarette consumption in the intervention group was one-quarter of that in the control group and over 10 years the net reported reduction averaged 53%. However, the normal-care group also reduced their consumption, reecting a general decline in smoking in the population, thereby reducing the apparent benet of smoking cessation in the intervention group over the ensuing years. Over the rst 10 years, the intervention group experienced fewer respiratory symptoms and less loss of ventilatory function; their mortality from coronary heart disease was 18% lower than controls, and for lung cancer it was 23% lower. No further contact with subjects to determine changes in smoking habits has been attempted, but follow-up has been continued for a further 10 years based on

RCT of antismoking advice: nal 20 years results (Rose G, Colwell L. J Epidemiol Community
Health 1992; 46: 757) Many studies have shown that the mortality and morbidity of ex-smokers is less than that of those who continue to smoke. On this basis, smokers are

Table 7.2 Neural tube defects (NTDs) and folic acid supplementation.

Folic acid + + -

Other vitamins + +

NTD/all babies 2/298 4/295 13/300 8/302

RR

} }

6/593 (1.0%) 0.28 21/602 (3.5%)

49

Chapter 7 Intervention studies death certicates and cancer registrations. Comparing the intervention with the normal-care group, total mortality was 7% lower, fatal coronary heart disease was 13% lower and lung cancer cases (deaths and registrations) were 11% lower. It was concluded that the policy of encouraging smokers to give up the habit was worthwhile and should not be changed. It was estimated that out of every 100 men who stopped smoking, between 6 and 10 were in consequence alive 20 years later. credence by the results of two intervention studies, both of which had methodological weaknesses. A large RCT was needed to resolve the matter. The trial was conducted in 33 centres in seven countries amongst 1817 women known to be at high risk through having had a previous affected pregnancy. They were allocated at random to one of four groups who received supplementation with folic acid and/or other vitamins or none. Of 27 babies born to these women with a neural tube defect, six were in the group who received folic acid supplementation and 21 in the other two groups a 72% protective effect (RR = 0.28) (Table 7.2). The other vitamins showed no benet. It was concluded that folic acid supplementation starting before pregnancy can now be rmly recommended for all women who have had an affected pregnancy. There are also grounds for public health action to ensure that the diet of all women who may bear children contains an adequate amount of folic acid.

Prevention of neural tube defects: results of the MRC Vitamin Study (MRC Vitamin
Study Research Group. Lancet 1991; 338: 1317) It has long been suspected that diet has a role in the causation of neural tube defects, one of the commonest severe congenital malformations. The possibility that supplementation with folic acid or other vitamins might reduce the risks was given

50

Chapter 8 Health information and sources of data

Introduction
Health is an elusive concept. The World Health Organization (WHO) denes it as a state of complete physical, mental and social well-being. It is, however, difcult if not impossible to use this denition to measure the health of populations in any categorical sense. Its principal limitation is that an individuals sense of well-being is intimately related to that persons expectations from life; these are difcult to measure objectively. Therefore, in order to measure and compare the health of populations, there are few alternatives other than to make use of indices of death and disease, despite the fact that these are the antithesis of health. The calculation of death rates and disease rates requires both numerator data about the events being studied (death and disease) and denominator data about the populations in which the events take place. This chapter is concerned with routinely collected data that can be used in the measurement of health, mainly from ofcial sources. Sources of data on the web
Ofce of National Statistics Census Health Protection Agency Association of Public Health Observatories Cancer Registration Census information http://www.statistics.gov.uk/ http://www.hpa.org.uk/ http://www.pho.org.uk/ http://www.ociu.org.uk/ links.html#Cancer_Registries www.census.ac.uk

Census data
Most developed countries undertake regular and detailed censuses of their populations in order to provide information to assist in social and scal planning. Although there is evidence that many of the ancient empires, for example Babylon and Egypt, undertook occasional, quite sophisticated censuses, it was the Romans who introduced it as a regular administrative exercise. They did this primarily for taxation assessment purposes. Perhaps the most famous Roman census was the one which took the parents of Christ to Bethlehem at the time of his birth. After the fall of the Roman Empire, the regular counting of populations ceased. In England, the rst post-Roman attempt to enumerate the population resulted in the compilation of the Domesday Book in the 11th century. In common with most of their predecessors, the administrators at that time were concerned to identify families rather than individuals, and even families of different status were recorded differently. From the material that survives, it is not possible to derive a precise gure of the population at that time. The modern system of censuses was introduced in Europe during the late 18th and early 19th centuries. In England and Wales, the rst complete census was undertaken at the behest of Parliament in 1801. Since then there has been a full census every 10 years, with the exception of 1941. In recent years, 10% sample censuses have been under51

Chapter 8 Health information and sources of data taken midway between the full decennial censuses. The census is conducted by the Ofce of National Statistics for England and Wales, the General Register Ofce in Scotland and the Northern Ireland Statistics and Research Agency. Each census is undertaken only with the specic authority of Parliament. Any individual who refuses to cooperate is liable to prosecution. The precise information that is collected varies from census to census but it invariably includes age, gender, marital state, place of birth, occupation, number of children, usual place of residence and duration of present residence. In addition, the head of the household has to furnish details of the residence including its type, tenure, accommodation and facilities. In recent years, it has been the practice to ask for additional information from a sample of the population. All the information relating to individuals is condential, even within government departments. Before census day, ofcials deliver the appropriate census form to each household and residential institution in the country. They are collected by the same ofcial after census day. Census ofcers are available to help householders with any problems they may encounter. The data on the forms are analysed centrally. In the past, tabulations of census data have been published as books, some of the more detailed information only after a delay of several years because of the time required for analysis and printing. Since 1981 census material has been available both as books and on computerreadable media. Much of the material from the 2001 census is available on the web. Despite some problems arising from concealment or misreporting of census information, and slight underrecording because some people are not at a formal address on census night, modern censuses are regarded as being generally very accurate. Information on the 2001 census is available at www.census.ac.uk. ducting deaths and emigrants from numbers recorded in the census, and adding births and immigrants. At the same time, the age distribution of the people remaining is adjusted. These are known as intercensal estimates. Unfortunately errors occur which are compounded by the passage of time. The principal sources of error in the intercensal estimates arise through inadequate recording of immigration and emigration in terms of numbers, age and sex. Furthermore, there is no system for ascertaining the amount of internal migration (changes of residence within the country). Thus, the greater the time that has elapsed since a census, the less the precision of the estimate, especially estimates relating to small areas within the country. After a census, the gures for years since the last census are recalculated, taking account of the information provided by the new census. These are called postcensal estimates.

Population projections
For planning purposes, it is often essential to have some idea of the likely size and composition of the population in years to come. The essential difference between population estimates and population projections is that an estimate is based on knowledge of the births, deaths and migration that have happened, and a projection is based upon what is thought likely to happen. Therefore, assumptions have to be made about trends in mortality, birth rates and migration. These are arrived at by extrapolation of past trends. Unforeseen changes in, for example, fertility can invalidate the projections.

Vital events
General
Since the early 19th century, there has been a statutory requirement for all births, deaths and marriages in the UK to be registered. Before the Births, Marriages and Deaths Act (1839) most of the records that existed were kept by the ecclesiastical authorities. Since the vast majority of people at

Estimates of population between censuses

The size and demographic characteristics of the population in non-census years is estimated by de52

Health information and sources of data Chapter 8 that time were baptized in infancy, the numbers of baptisms recorded in the parish registers can be used as proxy indicators of the numbers of live births. There was no information on stillbirths. As most marriages were in church, marriage rates can also be computed from the parish records. Likewise, most of the population received Christian burials and therefore the fact of death was usually recorded. However, the presumed cause of death was of little interest to the ecclesiastical authorities and was not routinely noted. In the 17th century, Bills of Mortality were published for some large towns and cities. The best known are those compiled by John Graunt (Fig. 8.1). The cause of death was arrived at by paying lay searchers, normally women parishioners, to inspect the bodies and form an opinion. Whereas many of the common causes of death left stigmata that were plain for all to see, for example plague and smallpox, other diseases gave rise to less denite changes and there was doubtless considerable guesswork on the part of the searchers. When the secular authorities made the registration of vital events mandatory, a government department called the Registrar Generals Ofce was established to supervise the processing and collation of records, and to report to Parliament and other government departments. Dr William Farr was the rst medical statistician at the ofce. His meticulous and imaginative analyses of the data set the standards for the present sophisticated system for the registration, analysis and publication of vital events. Now the task of collating, analysing and publishing information relating to vital events is the responsibility of the Ofce of National Statistics (website: www.statistics. gov.uk).

Birth registration
Childs name, sex, date and place of birth Mothers name, place of birth and usual residence Fathers name (if known), place of birth and occupation

Not all of these data are entered on the birth certicate. Additional condential information is collected for statistical purposes. This includes mothers date of birth and fathers date of birth (if his name appears on the register). For legitimate births, the following additional information is required: date of parents marriage, number of previous marriages of the mother and number of children born in the present marriage, distinguishing those born dead from those born alive. If the child is stillborn, a certicate of cause of stillbirth has to be presented to the registrar. This certicate is similar to a death certicate and is issued either by a registered medical practitioner or by a state-certied midwife involved with the birth. Tabulations and analyses of birth data are published annually by the Ofce of National Statistics and are available on the web (www.statistics.gov.uk). They are used to study patterns of fertility and to assist in making population estimates and projections.

Deaths
The present regulations governing registration of deaths were set out in the Births and Deaths Registration Act (1968). The Act requires that: . . . in the case of the death of any person who has been attended during his last illness by a registered medical practitioner, that practitioner shall sign a certicate . . . stating to the best of his knowledge and belief the cause of death and shall forthwith deliver that certicate to the Registrar. The certicate that the doctor is required to complete and sign (Fig. 8.2) is one of cause (or causes) of death, not of fact, since the doctor is not obliged to inspect the body after death. After giving the deceaseds name, age, date and place of death and 53

Births
All births must be registered by one of the parents (or someone on their behalf) with the local Registrar of Births, Marriages and Deaths within 6 weeks of the event. Certain of the information required at this time is entered in the register and is available for public scrutiny. The following information is available for public scrutiny.

54

Chapter 8 Health information and sources of data

Figure 8.1 The General Bills of Mortality for London, 1641 and 1665.

Health information and sources of data Chapter 8 details of how far the death was investigated, the doctor is required to state the immediate cause of death. There is then space provided for him or her to record the antecedent causes (giving the underlying cause last) and any other signicant conditions that may have contributed to the death. As far as possible, the doctor should use generally accepted terminology, such as that set out in the ICD. The Registrar requires him or her to avoid the use of indenite and ambiguous terms such as heart failure or old age. The completion of the certicate is quite straightforward in the case of an individual who has died as a result of a well-dened disease that has been extensively investigated in life, for example death by bronchopneumonia due to carcinomatosis due to carcinoma of the bronchus, with chronic bronchitis as a signicant condition that contributed to death. However, in many circumstances the death certicate cannot be completed with such precision, for example in the case of an old person who has previously had a stroke, has diabetes, has chronic cardiac failure, is known to have bronchitis, has been bedridden for months and who is found dead in bed one morning. In such cases, the certied cause of death is an arbitrary opinion rather than a statement of fact. Generally, the precision of death certication tends to diminish with increasing age of the deceased. An informant, who is usually a close relative of the deceased, a person present at death, the person in charge of the institution in which the person died or the person responsible for the disposal of the body, must register the death with the Registrar as soon after death as possible. When doing so, he or she must give the following information: Date and place of death. Full name and sex of deceased. Maiden name of married woman. Date and place of birth of deceased. Occupation and usual address of deceased. The data above are recorded in the register. If the Registrar is satised that the particulars are in order and that there is no need to report the death to the Coroner, he or she will issue a death certicate and authority for burial. Death registration data are collated and analysed by the Ofce of National Statistics. The causes of death that are analysed are normally those given as the underlying cause rather than the immediate cause because the former is more informative and more useful for the study of disease in the community. In the rst example given above, the death would be classied as due to carcinoma of the bronchus for purposes of statistical analysis. The tables published by the ONS must be interpreted with this rule in mind. They do not necessarily provide a complete picture of mortality attributable in whole or part to specic causes. In certain circumstances, a normal death certicate cannot be issued. These are when there was no medical attendant during the last illness of the deceased, when it is suspected that the death resulted from unnatural causes, or when the death occurred before full recovery from a surgical operation or the administration of an anaesthetic. In such circumstances, the death must be reported to the Coroner by either the attending doctor, or the police or the Registrar. A coroner is a member of the judiciary and is bound by legal processes. He or she has to be legally qualied but not necessarily medically qualied, though some have both qualications. The Coroner investigates the death by enquiry, either directly or through his or her ofcers. The Coroner may order a postmortem examination and may hold an inquest, with or without a jury. Having established the cause of death to his or her satisfaction, the Coroner will then sign a death certicate. If the Coroner has reason to believe that death was caused by the unlawful action of another person, he or she is bound to forward the papers to the Director of Public Prosecutions. It should be noted that in these circumstances it is the Coroners job to establish the cause of death, not who caused it. If those responsible for the disposal of the body wish the deceased to be cremated, an additional certicate is required. The person wishing to have the body cremated has to complete part of a form. The practitioner who attended the deceased during the last illness completes another part. This part has certain similarities to a certicate of cause of death but the doctor must have inspected the body 55

Chapter 8 Health information and sources of data

Figure 8.2 Death certicate (England and Wales). (Reproduced with permission of the ONS (Crown copyright).)

after death. The third part is completed by another medical practitioner who is not professionally associated with the attending practitioner nor related to him or her or to the deceased. He or she must have been on the Medical Register for at least 5 years. This second doctor must inspect the body and form the view that the cause of death is as stated by the other practitioner. The nal part is completed by the medical referee of crematoria for the local government authority involved. He or she has to afrm that the particulars on the other parts of the form are reasonable and have been completed by properly qualied doctors.

WHO denitions
Neonatal death Early neonatal death Late neonatal death A live-born infant that dies within 28 days A live-born infant that dies within 7 days A live-born infant that dies after 7 days but within 28 days A fetus that dies before birth but after a presumed 24 weeks of gestation A combination of stillbirths and early neonatal deaths Deaths from 1 month to 1 year of age Deaths under 1 year of age

Stillbirth

Perinatal death Postneonatal deaths Infant deaths

Stillbirths and infant deaths

Epidemiologists are particularly interested in the rate of stillbirths and infant deaths because they are a sensitive indicator of the general health of the population and also reect the quality of child health services. Comparison of death rates between countries and the associated trends over time are of special interest. So that such comparisons can be made, agreed denitions and terminology have been promulgated by the WHO (see box). 56

The Stillbirth (Denition) Act (1992) reduced from 28 weeks to 24 weeks the minimal gestational age by which a stillbirth is dened. This reclassication led to an apparent increase in the stillbirth and perinatal death rates. The infant death rate in the UK has fallen from around 150 per 1000 live births in 1900 to around 6 per 1000 in 2001. The current very low rate limits the possibility of

Health information and sources of data Chapter 8

Figure 8.2 Continued

further improvements. The commonest causes of neonatal death include congenital abnormalities and prematurity. Many of these deaths would seem to be unavoidable, although congenital rubella is one example which it is hoped will be completely eliminated through immunization of children. Around 40% of postneonatal deaths are due to cot death (sometimes called sudden infant death syndrome). Associations have been shown with maternal smoking, prone sleeping position, bottle feeding and season of the year. A campaign to encourage mothers to place their baby on their side or back rather than prone has led to a reduction in the number of cot deaths. The effectiveness of stopping smoking and encouraging breast feeding in reducing the number of deaths from cot death has yet to be shown. A small but increasing proportion of deaths in the postneonatal period are due to congenital abnormalities and conditions originating in the perinatal period, suggesting that some infants that previously died soon after birth are now living until the postneonatal period.

Abortions
A spontaneous abortion is the expulsion of the

product of conception before it has reached an age when it could be expected to have an independent life and shows no signs of life at birth. The lower limit of fetal viability is dened as 24 weeks gestation. Since the Abortion Act (1967) came into force, it has been permissible for a pregnancy to be terminated provided it has not progressed beyond 28 weeks gestation, if two doctors believe that the continuation of the pregnancy would be injurious to the physical or mental health of the woman or that there is a risk that the child may be born with a disability that would prevent it from leading a normal life. Under 1990 legislation, abortion is normally permissible only up to 24 weeks gestation. When a termination of pregnancy is carried out under the provision of the Act (it is illegal to terminate a pregnancy other than for reasons set out in the Act) the doctors involved have a statutory obligation to notify the Department of Health (DoH). The form of notication asks for the name, date of birth and marital status of the woman, her normal place of residence, and the number of previous pregnancies, distinguishing those that proceeded to term from those that were terminated. The presumed duration of the pregnancy, the statutory grounds for the operation and the place 57

Chapter 8 Health information and sources of data where it was carried out are also required. The forms are sent in condence to the DoH where they are checked to ensure that the law is not being abused. They are collated and analysed by the Ofce of National Statistics, which publishes annual tabulations setting out the number of abortions by different criteria. in most people are minor, self-limiting conditions for which there is no specic treatment. Diagnostic precision is unnecessary and it is a waste of time to attempt to discriminate between the many causes by complicated and expensive viral studies and other examinations. In such circumstances the data that are generated may not have sufcient precision for epidemiological studies. In many cases, the stage at which disease is treated depends on a complex series of factors other than the patients perception of the problem. These include the availability of health service treatments, for example waiting lists, outpatient appointment availability and the acceptability of the treatments that the patient believes will be offered. This will affect the morbidity recorded at hospitals and employment sickness absence gures. Finally, there is a quite separate problem in the way morbidity statistics are calculated and presented. The calculation of mortality rates is relatively straightforward because each individual can only die once. Thus, if there are 10 deaths in a population of 162, the death rate is 61.7 per 1000. If, however, 10 episodes of an illness occur amongst 162 people during a year it does not mean that 61.7 per 1000 population were ill one individual may have had more than one episode of illness; indeed, all the episodes may have occurred in the same individual. Many morbidity statistics are collected in such a way that it is impossible to distinguish episodes of illness from sick individuals. When presenting or making use of rates it is important to be clear how the rate was derived. Morbidity statistics routinely available in England and Wales include the following.

Morbidity
General
Morbidity statistics are concerned with the amount and types of illness that occur in the community. The sources of available data vary from place to place and from time to time. They include, for example, attendances for primary care, hospital outpatients and admissions, as well as statutory sources and special registers for particular conditions. Most routinely collected morbidity data suffer from serious shortcomings partly because of the ephemeral nature and imprecise diagnosis of many illnesses and partly because of inadequacies in the information systems. Consequently, although they should give a more complete picture of the incidence of disease in communities than mortality data, they do so with varying reliability and must be interpreted with caution. One of the principal problems centres around the denition of illness itself. For some people, a common cold or backache may represent an illness and justify them seeking medical help or being away from work. These peoples illnesses may be recorded in one of the many routine data systems. For other people, symptoms that the medical profession would regard as indicative or diagnostic of major disease may be regarded as having no serious signicance, an inconvenience to be tolerated until normal recovery takes place. Such illnesses will not feature in any morbidity statistics because those affected do not seek medical aid nor allow the symptoms to alter their lifestyle. Another problem is that diagnostic precision varies between doctors according to their perception of the disease that they are treating. For example, inuenza and upper respiratory viral infections 58

Morbidity statistics
Hospital episode statistics General practice databases Infectious diseases Notication of episodes of STDs Notication of prescribed and other industrial disease and accidents Notication of congenital malformations Cancer registration Laboratory reports on infections

Health information and sources of data Chapter 8

Table 8.1 Statutorily notiable infectious diseases. Under the Public Health (Control of Disease) Act (1984) Cholera Smallpox Plague Typhus Relapsing fever Under the Public Health (Infectious Diseases) Regulations (1988) Acute encephalitis Ophthalmia Acute poliomyelitis neonatorum Anthrax Paratyphoid fever Diphtheria Rabies Dysentery (amoebic or bacillary) Rubella Leprosy Scarlet fever Leptospirosis Tetanus Malaria Tuberculosis Measles Typhoid fever Meningitis Viral haemorrhagic Meningococcal septicaemia fever (without meningitis) Viral hepatitis Mumps Whooping cough Yellow fever

Hospital episode statistics

Hospital episode statistics (HES) provide data on hospital admissions (including day cases) in England. For each episode the age, gender and ethnic group of the patient is recorded together with other information (e.g. length of stay, waiting time, type of admission, specialty, diagnosis, operations). The system has been operating since 1989. Analyses are published annually and some tables are available on the web. Special analyses can be undertaken, subject to ethical approval. The website is http://www.doh.gov.uk/hes/.

General practice databases

For some years most general practices have been using computers in their practices to assist in the management of the practice and to maintain clinical records. There are many packages that are available. All of them have the facility to generate prescriptions and maintain les of demographic data, diagnoses and symptom description and results of investigations. In the late 1980s systems were developed to harvest data from consenting practices in order to create databases that could be used for research and other purposes. The data abstracted from the general practices is anonymized in order to ensure patient condentiality. The main systems now available are the General Practice Research Database (GPRD) (owned by the MHRA) and Mediplus (owned by Intercontinental Medical Statistics).

Infectious diseases
When a doctor suspects that a patient is suffering from a notiable infectious disease or from food poisoning (Table 8.1) the law requires him or her to send a certicate to the Proper Ofcer designated by the local authority. The ofcer responsible for infectious disease control is usually the local Consultant in Communicable Disease Controls (CCDC) employed by the Health Protection Agency (HPA). When it is a disease that is likely to require urgent control measures to be taken, the doctor will normally notify the CCDC by tele-

phone and provide the formal certicate later. Similar action will usually be taken in the case of nonnotiable infectious diseases (or outbreaks due to other causes, e.g. chemical poisoning) which may require immediate investigation. This is usually undertaken by the environmental health ofcer employed by the local authority. The importance of complete and prompt notication is not universally appreciated and therefore many infectious diseases are under-reported. Notication is important for a variety of purposes. In the case of some infections, such as food- and water-borne disease (food poisoning, typhoid, etc.), bacterial meningitis (particularly meningococcal infection), infectious hepatitis, diphtheria and tuberculosis, immediate action may be required to limit the spread of infection and to safeguard public health. Notications are also of value in studying the aetiological factors inuencing the incidence of disease in the community and in monitoring the effectiveness of vaccination and immunization and other programmes. 59

Chapter 8 Health information and sources of data in 1974. Doctors are required to inform the Proper Ofcer, who in this instance may be the CEHO (Chief Environmental Health Ofcer) of the local authority or the CCDC, of the occurrence of any of the notiable diseases they list. Doctors must also report poisoning by the following substances.

Notications of episodes of sexually transmitted diseases (STDs)

Genitourinary medicine clinics of the NHS are required to make regular returns to the Department of Health for England (or its equivalent in other countries of the UK) of the numbers of new attendances with STDs. The following are dened as STDs for this purpose.

Industrial poison reporting

Aniline Arsenic Benzene (chronic) Beryllium Cadmium Carbon bisulphide Compressed air Chrome Lead Manganese mercury Phosphorus

Sexually transmitted diseases

Syphilis Gonorrhoea Non-specic genital infection Trichomoniasis Candidiasis Scabies Pediculosis pubis Herpes simplex Condylomata acuminata Molluscum contagiosum Chancroid Lymphogranuloma venereum Granuloma inguinale Other attendances requiring treatment

Also cases of epitheliomatosis, toxic anaemia and toxic jaundice

For two of the above diseases, syphilis and gonorrhoea, the age of the patient and whether the disease was contracted outside the country has to be stated. In no case are data given that could identify an individual. Although this system provides a useful picture of the overall trends in STD it has to be appreciated that not all cases treated are seen in NHS clinics. The nature of the information that is collected means that it is of limited value for all but the most basic of epidemiological studies. In particular, it is not possible to distinguish episodes from numbers of people involved. Notications of HIV infection or AIDS are not included in clinic reports. Details of cases of AIDS are reported separately in condence to the Director of the CDSC (Communicable Disease Surveillance Centre) of the Health Protection Agency, which also collates reports from laboratories on HIV antibody-positive cases.

The HSE publishes the number of reported cases annually. Although the system is of great value in controlling these diseases, it undoubtedly gives an underestimate of the true incidence of these conditions. Many of the diseases occur in factories and work places in which there is no medical ofcer and, even if detected by another doctor, they are not always reported. Employers also have an obligation to notify the HSE of accidents (both fatal and non-fatal) which occur in their factories or work place. These are published in the annual report. A third source of data relating to industrial diseases is notications of prescribed occupational diseases, for example pneumoconiosis in coal miners, tuberculosis in medical laboratory workers and mesothelioma in asbestos workers. Workers who have these diseases are entitled to compensation under current regulations. Thus, notication is probably more complete than for many other diseases since there is a potential advantage in doing so to the individual with the disease. There are currently about 50 prescribed occupational diseases.

Industrial diseases and accidents

In order to improve personal safety at places of work, the HSE was established as a statutory body 60

Congenital malformations
A national scheme for the notication of congenital malformations was instituted in England and

Health information and sources of data Chapter 8 Wales in 1961 after an episode in which thalidomide was responsible for a major outbreak of limb deformities in the children of mothers who had taken the drug during early pregnancy. There is no statutory requirement on doctors or midwives to notify cases. One of the problems with these data is the denition of malformations. There is little difculty in detecting a major malformation but some minor abnormalities may not be noticed, or if noticed are not deemed to be of sufcient importance to justify notication. main purposes are to provide accurate knowledge of the incidence and prevalence of disease in a community which will assist in the organization and monitoring of its health services and in disease surveillance activities. Ideally, every health event and every kind of health resource would be recorded in a systematic and instantly available form. In practice, this is neither possible nor desirable as it would require an enormously complex and expensive system which would be too slow and cumbersome to be of value. Most health information systems have been developed to meet particular needs. Nevertheless, the data are often inaccurate and the system does not always allow users questions to be answered with ease. These shortcomings tend to bring systems into disrepute and the enthusiasm for collecting data (as well as making use of it) wanes. Some of the commonly encountered problems of information systems are as follows.

Cancer registration
Malignant disease has long been a major cause of morbidity and mortality in the UK and in most other countries. In order to study these diseases, it is essential to know the numbers of people affected by different forms of cancer and their survival rates. The system of cancer registration was set up in 1971 specically to facilitate research in this eld. Hospitals notify new cases of cancer to their regional centre. Data from all regions are analysed further by ONS. Periodic ofcial reports are published giving detailed tabulations of incidence, survival and mortality rates for various malignant diseases at different stages.

Information system problems

Lack of motivation among recorders Design of data capture procedure Inexibility in the system Irrelevance of analyses

Laboratory reports
The CDSC of the PHLS receives weekly reports from microbiology laboratories in England and Wales on cases of laboratory-diagnosed infections. The amount of clinical and epidemiological data reported varies depending on the infection. Although the data are incomplete and lack denominators which prevents their use to calculate incidence rates, they provide a useful means of monitoring trends and detecting outbreaks. The CDSC also collects data related to infectious disease from other sources, for example reports of outbreak investigations and immunization statistics.

Lack of motivation among recorders Often a low priority is accorded to the task of record keeping. This leads to delays in completion and poor quality of records, for example inaccurate information, items missing or no record at all. For these reasons all arrangements for data capture, as it is called, should be simple to operate and create the minimum amount of work. Design of data capture procedure The type of record needed for an information system is not always compatible with that required for clinical purposes. It is often possible, however, to use records made for other purposes if they are carefully designed, i.e. in standard format with provision for coding, etc. This requires a degree of collaboration between different interests which it is often hard to achieve. 61

Health information systems

Information systems are used to assemble facts and gures from a variety of sources for analysis. Their

Chapter 8 Health information and sources of data Inexibility in the system The need for simplicity in records means that the number of items recorded has to be restricted. Some exibility can be gained by allowing room on the record for additional items of local interest beyond a set of basic data required of all recorders. Irrelevance of analyses Users may feel that standard analyses tell them nothing new or are unhelpful in solving their problems. This tends to sap enthusiasm for the system. In the design of a routine information system, therefore, the following requirements should be met. The intended uses of the system should be specied so that the data recorded will be appropriate to their purpose and the collection of irrelevant data can be avoided. The recording procedures should be standardized and the data collected should be easy to obtain, accurate and as complete as possible in order that reliable comparisons can be made over periods of time and between different places. Data should be collected from all relevant sources for collation and analysis at a central point. There should be well-organized provision for data storage, updating, processing and retrieval. The system should be capable of providing answers to enquiries within the eld for which it is designed, with speed and accuracy.

62

Chapter 9 Indices of health and disease, and standardization of rates

Introduction
The health of a community can be measured by the appropriate use of basic morbidity and mortality data, including those from sources described in the previous chapter. To be intelligible, however, the crude numbers need to be presented in a form that allows valid comparisons to be made between groups, between years and between different areas. There are certain conventions in the handling and presentation of data. There are four main types of variable as shown below: qualitative (categorical) ordinal continuous quantitative discrete quantitative.

Ordinal variables
These are used to rank the quality of characteristics in order of severity, importance, etc.; for example, pain might be classied as 0 = none, 1 = some pain, 2 = severe pain, 3 = very severe pain. The analysis of such variables requires different statistical techniques to quantitative variables.

Continuous quantitative variables

These measure attributes that can occur at any point on a scale, for example height, weight or blood pressure. The degree of precision to which a continuous variable is measured depends upon its intended use in a particular investigation and the discriminatory power of the measuring instrument.

Types of variable
Qualitative variables
These are descriptive of a xed attribute, for example gender, religion, occupation and nationality. Such data are sometimes labelled for convenience by using numbers, e.g. 1 = male, 2 = female; or 1 = Church of England, 2 = Roman Catholic, 3 = Methodist, etc. These numbers have no meaning other than as labels.

Discrete quantitative variables

These measure attributes that can occur only as whole numbers (integers), for example the number of children born to a woman or the number of deaths in a year.

Grouping of data
For convenience of handling and presentation, continuous variables may be grouped as if they were discrete. For example, body mass index (BMI) 63

Chapter 9 Indices of health and disease could be grouped as less than 20, 2024.9, 2529.9, 3034.9 and 35 and over. The denition of groups is determined by the purpose of the investigation and the characteristics of the population being investigated. Thus the categories of weight used in studies of infants will be quite different to those used in adults. Discrete variables may also be grouped to produce larger numbers in each category. The class intervals between successive groups should usually be equal but it is often convenient to group all values at the extreme ends of a scale, which it must be remembered distorts the frequency distribution. Situations in which groupings are natural should be distinguished from those where they are arbitrary; for example, under 16 years and 16 years and over could be regarded as natural groupings in as much as people in the former category cannot be married and those in the latter can. For other variables, for example blood pressure, there is no such natural division. It is possible arbitrarily to dene systolic blood pressure in excess of 140 mmHg as high and below that level as not high, but this does not necessarily have any signicance. Quantitative data rarely fall into natural categories. Age-specic rates relate the number of events in people in a specied age group to the total population in that age group, for example y deaths per 1000 men aged 4564 years per year. Cause-specic rates relate cases of a specied disease to the population at risk, for example z cases of stroke per 1000 hypertensive patients per year. Such rates must always have a specied time dimension.

Special events related to total events

Examples Stillbirths are usually expressed as x per 1000 total births. Operative mortality can be expressed as y deaths per 1000 operations. Case fatality rates relate the number of deaths from a particular illness to the total number of cases of that illness. These types of rate are not time dimensioned but the period over which they were calculated should always be specied.

Incidence and prevalence rates Rates

It is rarely useful to state numbers of events alone. These can be interpreted only when they are related to a denominator, i.e. expressed as a rate; for example, it is not helpful to say that the number of deaths from pneumoconiosis is greater in coal miners than in, say, farm workers without relating the gures to the numbers of people employed in the two occupations. Two types of rate are frequently used: rstly events related to a population and secondly special events related to total events. In order to demonstrate how incidence and prevalence rates are derived, the mortality and morbidity experience of the employees in a hypothetical factory is shown schematically in Fig. 9.1.

Prevalence rates
The prevalence of a disease is the number of cases of that disease in a dened population at a particular point in time (point prevalence) or during a specied period (period prevalence). Both are expressed as a rate (x per 1000 population). From Fig. 9.1, the point prevalence rate at the beginning of the year was four per 18 people and at the beginning of August it was two per 18 people (one had died and one had left since the beginning of the year and two had joined). The period prevalence for the year was 12 per 18 people (by convention the denominator is the mid-year population). The

Events related to the population

Examples Birth rates are usually given as x per 1000 total population per year.
64

Indices of health and disease Chapter 9

D Died J Joined L Left

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Year A

Subject number

D L J D L Year B Year C

Figure 9.1 Morbidity and mortality experienced in a hypothetical factory.

period prevalence approximates to the sum of the point prevalence at the beginning of the period and the incidence during the period.

Cumulative incidence
Another method of expressing incidence is cumulative incidence which takes the number of people at the beginning of the study period as the denominator and the number of events that occur in that population as the numerator. It is a useful measure when estimating the likelihood that someone will contract or die from a particular disease.

Incidence rates
The incidence of a disease or other events is the number of new cases that occur during a specied period in a dened population. The most accurate way to estimate this is by calculating the denominator of each individuals time at risk. This can be approximated as the number of people in the population at the mid-point in time multiplied by the period of the study. Thus, from Fig. 9.1, in year B, the incidence of illness was 8. (The rst illness in subject 11 and the illnesses in subjects 14, 19 and 20 started before the beginning of the period specied and are therefore discounted.) The incidence rate is 8 per 18 people per year, i.e. the mid-year population is used as the denominator. It should be noted that the incidence for a specic period is only valid for that period. Thus, in the 6 months JanuaryJune of year B, the incidence of disease was 5; it is clear that it cannot be multiplied to give an incidence of 10 during a 12-month period.

Case fatality rate

An important concept, which is similar to that of cumulative incidence, is case fatality. Here the denominator is the number of people with the disease and the numerator is the number dying from that disease. The period at risk does not need to be specied but for some diseases such as meningococcal meningitis or myocardial infarction it is usually relatively short. Case fatality rates for cancers are measured over longer periods.

Error in health information

The value of data ultimately depends on how accurately they reect the true frequency of the disease 65

ar ch Ap ril M ay Ju ne Ju ly Au g Se pt Oc t No v De c Ja n Fe b

Oc t No v De c Ja n Fe b

Chapter 9 Indices of health and disease (or other variable being measured) in the population concerned. This section sets out some of the common sources of error that may affect routine health information and the steps which can be taken to reduce their effects. Errors affecting mortality and morbidity rates are of two kinds, as shown below. tory; cultural and social background; occupation; economic constraints (e.g. paid sick leave); and availability of medical care (which is related to numbers of doctors, distance from doctors surgery, number of hospital beds and appointments systems). The effect of variations in illness behaviour is most marked in mild, non-fatal and self-limiting conditions.

Errors affecting mortality and morbidity data

Those affecting the numerator, e.g. diagnostic inaccuracy incomplete identication of cases variability of the recording system Those affecting the denominator, e.g. population migration changes in population structure changes in administrative boundaries

The recording system

The completeness and comparability of different sources of data may be affected by: the doctors view of the value of records; the simplicity and efciency of a records system; and changes in the conventions for coding and classication of disease or rules for selecting priorities among multiple diagnoses.

Denominator error Numerator error

The number of recorded cases of a particular disease may be in error for many reasons including the following. The size of population at risk often cannot be dened accurately and various methods of estimation have to be used. Some reasons for this are: population migration between censuses which may increase or decrease the size of population within an area; changes in population structure within different areas (e.g. age, race, occupational distribution), due to migration, changing fertility patterns, housing and industrial decay or development; and changes in administrative boundaries for reasons that may or may not relate to health and the provision of health services.

Diagnostic inaccuracy
This is affected by: the training, skills and interests of the attending physician; advances in medical knowledge of pathogenesis; variations in the criteria accepted in dening a diagnosis; and differences in the availability and use of special investigations. For example, until the mid-20th century carcinoma of the cervix was not distinguished from carcinoma of the endometrium in routine mortality statistics both were classied as carcinoma of the uterus. It is important to be aware of changes in the precision of diagnosis and classication when investigating time trends in the incidence of disease.

Reduction of error
The effects of errors such as those above can be reduced as follows. By use of a standard diagnostic classication such as the ICD when recording mortality or modications of this for morbidity. By combination of diagnostic categories between which transposition of cases may occur, e.g. cancer of the colon and large bowel obstruction. By use of standard recording and registration procedures.

Incomplete identication of cases

The probability that patients will consult a doctor or be seen at or admitted to hospital, for example, is inuenced by such factors as: past medical his66

Indices of health and disease Chapter 9 By use of denominator populations derived from similar sources and compiled by comparable procedures. Errors in routine statistics can rarely be completely eliminated. Therefore, caution is needed in their interpretation, particularly between different localities and at different times (see also Chapter 4). occurred in a designated group with that of a standard population. It is the ratio (usually expressed as a percentage) of the number of deaths which occurred in the designated group to the number that would have been expected if the mortality rates in each age band of the designated group had been the same as those of the standard population. Thus, the death rates for each age and sex group in the standard population (Mx) are multiplied by the number of people of that age and sex in the population being investigated (Px). This gives the expected number of deaths in that particular age/sex group. The expected deaths for each age/sex group are then added to give the expected number of deaths in the whole population being investigated. The observed number of deaths (D) is then divided by the expected deaths to give the SMR: SMR = observed deaths ( D ) 100 expected deaths [ ( Px M x )]

Standardization of rates
Rates calculated by using the total number of events as the numerator and the total population as the denominator are called crude rates. Their value is limited, particularly when comparing two populations with different age structures, for example mortality rates in a new housing development with many young families and those in a coastal resort with a large retired population. In these circumstances, it is essential to adjust the data to take account of the age differences between the populations; this is called age standardization. The two methods of standardization most frequently used are indirect standardization and direct standardization.

Indirect standardization
A commonly used method of indirect standardization for age is to calculate the standardized mortality rate (SMR). The SMR compares the mortality (either from a specic disease or for all causes) which

Example Members of the armed forces tend to be younger than the male population of the country as a whole. Therefore, the fact that they have a lower mortality rate is not illuminating. It is necessary to examine the mortality of this occupational group after taking account of the age factor. Their SMR for ischaemic heart disease (IHD) is calculated in Table 9.1. This indicates that mortality from IHD

Table 9.1 Mortality from ischaemic heart disease (IHD) in men serving in the armed forces. Death rates from IHD in England and Wales (per 1000) (Mx) 0 0.06 0.50 2.01 6.05 Population of armed services (1000s) (Px) 165.03 73.24 42.25 15.93 4.67

Age group (years) 1524 2534 3544 4554 5564 Total SMR =

Expected deaths (Mx Px) 0 4.39 21.13 32.02 28.75 85.79

Observed deaths 1 6 22 43 76 148

observed 148 100 100 = = 173 expected 85.79

Note: SMRs for occupational subgroups are usually conned to people aged 1564 years because the working population is conned to this age group.

67

Chapter 9 Indices of health and disease amongst men in the armed forces after adjusting for age distribution is higher than the national experience by a factor of 1.73. Another example of how standardization can be helpful is in comparing mortality in different years. The age structure of the population of England and Wales has been changing for many years and therefore crude death rates can give a misleading impression of changes in mortality. The SMR gives a clearer indication of the true picture (Table 9.2). This indicates that mortality in males in England and Wales declined between 1965 and 1973.

Direct standardization
Direct standardization for age involves calculating the age-specic death rates in the study population and applying them to the same age groups in a standard population. This can be real or hypothetical. In this way, the number of deaths that would have occurred in the standard population, had it experienced the same death rates as the study population, can be computed and compared with other groups. The method of direct standardization is shown in Table 9.3.

Table 9.2 Indirect standardization: mortality in males in England and Wales in 1965 compared with 1973. Death rate per 1000 males in England and Wales, 1965 (Mx) 21.8 0.9 0.5 1.0 1.1 2.5 7.4 21.4 53.0 118.4 242.4 Male population in England and Wales (1000s), 1973 (Px) 355.3 1561.7 4037.3 3534.2 3337.5 2877.6 3033.6 2643.1 1855.9 639.7 112.8

Age group (years) <1 14 514 1524 2534 3544 4554 5564 6574 75+ 85+ Total SMR =

Expected deaths (Px Mx) 7746.0 1406.0 2019.0 3534.0 3671.0 7194.0 22 449.0 56 562.0 98 363.0 75 740.0 27 343.0 306 026.0

Observed deaths

296 546

observed 296 546 100 = 100 = 97 expected 306 026

Table 9.3 Direct standardization: standardization of mortality in England and Wales in 1949 against the 1979 population. The age-standardized 1949 death rate (against the 1979 population) is 339 396.5/24 002 = 14.14 per 1000. This can be compared directly with the crude death rate for 1979 which was 12.41 per 1000. Age group Population, 1949 (a) (1000s) 3417 2869 3339 3189 3178 2335 1727 957 228 21 239 Deaths, 1949 (b) 17 643 2345 5031 6839 16 062 32 097 60 580 77 127 42 218 260 278 Death rate, 1949 (b/a) 5.16 0.82 1.51 2.14 5.05 13.75 35.08 80.59 185.17 Population, 1979 (c) (1000s) 3339 4063 3534 3326 2020 2924 2257 1384 355 24 002 Expected (c b/a) 17 231.3 3331.7 5336.3 7117.6 14 241.0 40 205.0 74 661.6 111 536.6 65 735.4 339 396.5

09 1019 2029 3039 4049 5059 6069 7079 80+ Total

68

Chapter 10 Medical demography

Introduction
Despite the presence of many serious endemic diseases and the occurrence of major epidemics and wars, the populations of most European countries increased substantially between the 16th and 20th centuries. In recent years there has been a reduction in the rate of growth of populations in many European countries. It seems likely that in the foreseeable future either they will remain stable or there might even be a modest decrease. The growth of the European resident population since the 17th century underestimates both the extent to which the numbers of European people increased and the rate at which the increase took place. Over several hundred years people emigrated in large numbers mainly to the Americas, Australasia and parts of Africa. The migrations were prompted by economic hardship, social pressures and religious persecution as well as for trading reasons and fortune hunting. A large proportion of the present populations of North America and Australasia are descendants of these migrants. Their numbers now exceed those of the parent (European) populations. Whether or not the European population would have increased in size to the extent that it has without migration and dispersal throughout the world can only be a matter of speculation. It is unlikely that it would have done, as the natural resources of

Europe would have been insufcient to support so large a population. Furthermore, the technology to create a safe urban environment, with pure water, adequate sanitation and means for the bulk transport of food, did not exist until recently. The populations of most other parts of the world began to increase more recently and their rate of increase has reached that prevailing in Europe in the 18th and 19th centuries only during the last century. An important difference between the contemporary situation in many of the poorer developing countries of the world and Europe in previous centuries is that there are no longer large, sparsely populated continents rich in natural resources that can be colonized and in which people can thrive. Thus, population growth, which in previous generations was regarded as a national problem, is now a world problem. It is forecast that if the prevailing rates of growth are sustained, the worlds population, now about 6.4 billion people, will reach 9 billion by 2050. The earths mineral and energy resources are nite and the rate at which they are being consumed, particularly by the industrialized countries, is increasing. In many parts of the world, there is a hopeless inability to meet local needs, and resources are inadequate to enable them to import essential commodities. It is predicted that, unless there are reductions in both the rate of population growth and the rate at which natural resources are consumed, there will be a catastrophic failure to 69

Chapter 10 Medical demography meet the basic needs of the majority of humankind within the next few generations. Cataclysmic prophecies that humankinds future is threatened in this way are not new. They have been widely debated since the 18th century. Probably the best known writer associated with the problems of overpopulation is the Reverend Thomas Malthus, an 18th century English clergyman, who attracted attention by his essay on The principles of population as it affects the future improvement of society. The two principles from which he argued were: that food is necessary for the existence of man and that the passion between the sexes is necessary and will remain nearly in its present state. He argued that the power of the population to reproduce was greater than power of the earth to produce food. He concluded that there must be a strong and constantly operating check on population from the difculty of subsistence. This conclusion led him to recommend that there should be no extension of relief for the poor, as this would articially reduce the difculties of subsistence and lead to uncontrolled population growth! The time scale within which he predicted catastrophe was wrong, partly because he did not foresee emigration and colonization. His contention that difculties in subsistence would act as a constant check on population growth has also been proved wrong by the experience in the countries of Latin America, the Indian subcontinent and elsewhere. At about the same time as the ideas of Malthus were being debated in Europe, similar discussions were taking place in China. Hung Wang Chi noted in 1793 that during a long reign of peace the government cannot prevent people from multiplying themselves, yet its remedies are few. One of the solutions that he suggested was to legalize and encourage female infanticide. Discussions of the problems of population have continued throughout the world up to the present time but now more is known about the size of the world population, the dynamics of growth and the potential resources of the earth. The United Nations, through its various agencies, regards population growth as one of the major world problems that will affect the quality of life, health and survival of humankind. 70 The areas with high population growth are mainly developing countries where there are already regular famines, chronic poverty, frequent epidemics of crippling diseases and declining living standards. The situation will only be remedied if those countries with the highest growth rates in population achieve stability and the countries with the highest growth rates in consumption of resources reduce their demands. The global problem of population growth is compounded by the fact that people are not evenly distributed on the habitable surface of the earth. Food shortages and disease are problems in some areas simply because of the local density of population rather than because the area as a whole has insufcient natural resources. It is important to recognize that health depends as much upon the systems for the distribution of food and water and the disposal of waste as it does upon the quantity of food produced or the availability of medical services.

Populations and growth rates

The size of the worlds population and its growth rate is arrived at by collating data from every country. The quality of the data varies considerably from country to country. Most of the richer industrialized countries undertake regular and detailed censuses similar to those undertaken in England and Wales (see p. 51). They also have sophisticated and comprehensive systems for the registration of births, deaths and marriages. From these sources it is possible to build up a complete picture of the way in which the size and structure of the population changes. In the poorer countries of the world national censuses are conducted infrequently and tend to be incomplete. The additional data that are required for demographic studies (e.g. the registration of vital events) are often defective. There are particular difculties in the most deprived sections of these countries and amongst nomadic peoples or those living in sparsely populated regions of the world with poor communications. In these latter situations, much of the data are available only on an irregular sample basis. It is not surprising that

Medical demography Chapter 10 most of the work on population growth has used European data, because only in recent times has it been possible to study many of the other countries of the world. The trends in population growth in England and Wales are not dissimilar to those in most European countries and can be used to illustrate the size and speed at which changes occurred. It has proved possible to estimate the number of residents at various times between 1100 and the early 19th century from analysis of ecclesiastical and governmental records. From the 19th century onwards formal census gures are available. The trend has been for the population to increase exponentially (Fig. 10.1). The temporary decreases in population due to major national disasters such as epidemics of plague or war are not discernible within the scale used on the gure but at the time they had major impacts in some parts of the country. For example, Fig. 8.1 (p. 54) shows the General Bills of Mortality for London for 1641 and 1665. In both years, the number of deaths greatly exceeded the number of births, in 1665 by a factor of 10. It should be remembered that Fig. 10.1 is solely concerned with the resident population and that during much of the period there was substantial migration. It should also be noted that the scale of the gure is such that the recent reduction in population growth rate is not apparent. At the same time as the population increased, its age structure changed. Figure 10.2 compares the age distribution of the population in 1821 with that in 1991. In 1821, the proportion of children was much greater than at the present time and the proportion of people over the age of 50 was considerably less. The population can only increase if the number of births exceeds the number of deaths. The growth rate of human populations tends to be exponential because with each annual increase in births the proportion of the population potentially capable of reproduction increases. For this reason, the statement that there is an annual growth rate of x per 1000 population (x being the difference between the birth rate and the death rate) gives a misleading impression of the magnitude of change. The conventional way of expressing growth is the population doubling time. This is the theoretical period that it will take for a given population to double, based upon the most recently available data. Clearly, the doubling time will have to be

40 1 000 000s 30 20 10 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 Centuries Figure 10.1 The growth of population in England and Wales.

80+ 7079 6069 5059 4049 3039 2029 1019 09 30

1821

2001

Figure 10.2 Comparison of the age structure of the population of England and Wales in 1821 with that in 1991.

Age

20

10 Per cent

10

20

71

Chapter 10 Medical demography revised when there is a change in either birth or death rate. The doubling time for the population of the UK, together with that for a number of other countries, is given in Table 10.1. tive of whether they are both high or both low. This phase is represented in Fig. 10.3 as period A. Typically primitive rural societies and poorly developed urban societies tend to have high birth and high death rates. The highest mortality tends to be in infancy and childhood due to the combined effects of disease and poor nutrition. Social progress and the introduction of industrial technology bring tangible and immediate benets to the community. The most obvious are improvements in sanitation, in water supply and in the ability to distribute and store food. The immediate effect of these changes is that the chances

Demographic transition
The model of demographic transition provides a useful framework within which to consider the factors that determine changes in the size and structure of human populations. The population is stable both in size and in age structure when the birth and death rates are equal and static, irrespec-

Table 10.1 Population doubling times in various countries and regions of the world. (Source: WHO, 1991.) Population doubling time (years) 19.8 28.3 30.2 50.6 54.5 69.6 72.5 87.0 90.4 115.9 178.1 210.4 217.0 277.6 408.1 495.5 1386.6 1386.6

Country Kenya Cambodia Mexico Argentina Singapore New Zealand India Canada USA Ireland Russian Fed. Sweden Japan UK Spain Belgium Denmark Italy

Population (1000s) 25 905 5729 81 141 32 609 2705 3380 844 324 25 309 248 710 3523 143 585 8635 123 611 57 367 39 025 9844 5148 57 052

Birth rate (per 1000) 47.0 41.4 29.0 21.7 17.8 17.8 29.9 15.3 16.3 14.9 14.6 14.3 9.9 13.8 10.2 12.0 12.4 9.8

Death rate (per 1000) 11.3 16.6 5.8 7.9 5.0 7.8 20.3 7.3 8.6 8.9 10.7 11.0 6.7 11.3 8.5 10.6 11.9 9.3

Life expectancy, males (years) 56.5 47.0 62.1 65.5 68.7 71.9 55.4 73.0 71.8 71.0 64.2 74.8 75.9 72.4 73.2 70.0 71.8 73.2

Life expectancy, females (years) 60.5 49.9 66.0 72.7 74.0 78.0 55.7 79.8 78.6 76.7 74.5 80.4 81.8 78.0 79.1 76.8 77.7 79.7

Fertility 7.0 4.7 4.2 3.0 2.0 2.2 4.0 1.8 2.0 2.2 2.0 2.1 1.6 1.8 3.7 1.6 1.6 1.3

World region Africa Latin America Asia Oceania Former USSR North America Europe World

Population (millions) 662 457 3171 27 291 278 500 5385

Birth rate (per 1000) 45.0 29.0 28.0 19.0 18.0 15.0 13.0 27.0

Death rate (per 1000) 15.0 7.0 9.0 8.0 11.0 9.0 11.0 10.0

Population doubling time (years) 23.5 31.9 36.8 63.4 99.4 115.9 346.9 41.1

72

Medical demography Chapter 10 of survival amongst the most vulnerable within the community, infants and children, are improved. Therefore the death rate begins to fall and the community enters phase B in Fig. 10.3. During this phase, the crude birth rate actually rises because the proportion of the population that is capable of reproduction increases and there is little change in the age-specic birth rates. This is because peoples reproductive behaviour tends to be learned from their parents and it can take many years to adapt fully to new circumstances. In many societies, the desirability of large families, which is a biological necessity for survival in pretransitional communities, is formalized within the belief system of the group. For example, in many societies, the number of children a man has is perceived as a measure of his virility. The next phase (C in Fig. 10.3) is characterized by a decrease in the birth rate while the death rate continues to fall. Birth rates still exceed death rates and the exponential growth of the population, established in phase B, continues. Again this is because, despite a decrease in the average number of children born to each woman, there are more women in the reproductive age group than there were in the previous phase. Eventually death rates stabilize (phase D) but birth rates continue to fall. The transition of the society is completed in phase E, when birth and death rates are static and equal. By this time, the size of the population is many times greater than it was in the pretransitional phase. The size of the new stable population is determined by the speed of the transition. Data from England and Wales can be used to illustrate the demographic changes discussed above. The crude and the age-specic death rates for selected age groups relative to the 1841 rates in England and Wales are shown in Fig. 10.4. The crude

Birth rate

Population

Death rate

Figure 10.3 Schematic representation of demographic transition.

100 Death rate per 10-year period (percentages) 90 80 70 60 50 40 30 20 10 59 year olds

6574 year olds

Crude death rate

3544 year olds

Figure 10.4 Age-specic death rates per 10-year period for England and Wales since 1841, as a percentage of the 184150 age-specic rates.

1841 51 61 71 81 91 1901 11 21 31 41 51 61 71 Year

73

Chapter 10 Medical demography death rate is now about half what it was in the early 19th century. The greatest changes in mortality have been amongst the young, exemplied by the 59 year olds in the gure, which are now less than 5% of the rates prevailing in the early 19th century. The smallest changes have been amongst the elderly. This is reected in the change in life expectancy, another way of summarizing mortality, at different ages (Fig. 10.5). It is arrived at by applying the prevailing age- and sex-specic mortality rates to the people who survive to a particular age. It is clear that the greatest changes in life expectancy have been amongst the very young. Increased survival in the prereproductive age groups means that the proportion of the population capable of reproduction increases. Thus, although each age group of women may maintain the same agespecic fertility rates as previous generations, the crude birth rates will rise. ern world, owed more to improvements in the general quality of life and to improvements in public and personal hygiene than they did to any specic medical measures. Specic medical treatments were not introduced until long after the mortality rates from these diseases had undergone the greater part of their fall. It is noteworthy that many of the lethal diseases of 19th-century Europe are now regarded as tropical diseases. They are more properly called poverty diseases. The principal diseases that accounted for the high mortality and which have now been controlled or eliminated in the western world were tuberculosis, the enteric fevers, cholera, smallpox, scarlet fever, measles, whooping cough and diphtheria. During the 1840s, about 18% of all deaths in England and Wales were attributed to tuberculosis. It is possible that some of these may have been misdiagnosed carcinoma of the bronchus or some other disease of the respiratory system, but the numbers were so large that there can be little doubt that the downward trend in mortality rates shown in Fig. 3.1 (p. 15) was mainly a reection of tuberculosis control. The decline in tuberculosis mortality preceded the identication of the organism or any specic treatment. The principal explanation for this remarkable trend, however, probably lies in improvements in diet and in consequent enhancement of the resistance of individuals. The practice

Reasons for the decline in mortality

The reduction in mortality in England and Wales since the 19th century is almost entirely due to the elimination of the major endemic infectious diseases (Fig. 10.6). For most of these, mortality rates were highest amongst young people. It is apparent that the virtual disappearance of these diseases from the UK, and from most countries in the west-

Deaths per 1000 population

70 60 Expected years of life 50 40 30 20 10 0 10 20 30 40 Age 50 60 70 80 1841 1979

24 20 16 12 8 4 0 1838 39 40 41 42 Year

Infectious diseases Non-infectious diseases

1969 70 71 72 73

Figure 10.5 Expectation of life at different ages in England and Wales, 1841 and 1979.

Figure 10.6 Crude annual death rates from infectious and non-infectious diseases in England and Wales, 18381842 and 196973. (Death rates from infectious diseases during 196973 were too low to show on this scale.)

74

Medical demography Chapter 10 of isolating cases, thereby reducing the spread of the disease, probably also had an effect. The enteric and diarrhoeal diseases were endemic in the 19th century and were a particularly important cause of death amongst infants and children. Their impact began to decline in the mid19th century (Fig. 10.7) and seemed to be the result of improvements in personal hygiene and in childrearing practices. A more specic measure, the provision of a pure water supply, was responsible for the disappearance of cholera as an endemic disease in the UK (Fig. 10.8). Because of the obvious physical signs of smallpox, the statistics on its mortality are likely to have been accurate. This disease was endemic in the 19th century (Fig. 10.9). Typically, there were superimposed regular epidemics every 67 years. The frequency of these epidemics was probably due to changes in population immunity. Contact with disease resulted in either death or lifelong immunity, thereby reducing the size of the susceptible population. After an epidemic, most survivors would be immune and this decreased the risk to the remaining susceptibles. When the proportion of susceptibles in the population increased (by the birth of children), a further epidemic occurred. Not surprisingly, the majority of deaths occurred amongst children and infants. The elimination of this disease was due to a specic medical measure, the discovery of vaccination. However, it should be noted that although vaccination became compulsory in England in 1852, it was not widely practised for a further 20 years. Other infectious diseases that ceased to be a major cause of mortality included scarlet fever, which was endemic and had regular superimposed epidemics. Its eventual elimination could have been due to the advent of more successful treatments for the complications of the disease or to changes in the virulence of the organism. Many of the measures that have achieved the control of the infectious diseases are to a large extent by-products of improvements in the quality of life and, more recently, relatively simple medical measures. All should be applicable and are being applied in poorer countries of the world at the present time with consequent accelerating reductions in their mortality levels.

300 280 260 240 220 200 180 160 140 120 100 80 60 40 20 1851 55 59 63 67 71 75 79 83 87 Year

Figure 10.7 Mortality rates from enteric fevers, England and Wales, 185189.

Death rate per 100 000 population

Death rate per 100 000 population

300 280 260 240 220 200 180 160 140 120 100 80 60 40 20
1847 49 51 53 55 57 59 61 63 65 67 69 71 73 75

Death rate per 100 000 population

100 75 50 25
1851 53 55 57 59 61 63 65 67 69 71 73 75 77 79 81

Year
Figure 10.8 Cholera mortality in England and Wales, 184777.

Year Figure 10.9 Smallpox mortality in England and Wales, 185181.

75

Chapter 10 Medical demography age of marriage. It is, however, used as a means of reducing population growth in some countries, notably China. Although conception may still take place below the minimum age for marriage, the pregnancy is stigmatized as illegitimate. The conventional age of marriage tends to be several years greater than the legal minimum. In 19th-century Sweden, the conventional age of marriage was the middle to late twenties. This convention was imposed in rural communities by obliging a man to demonstrate his ability to support his wife before marriage could take place and by his living apart from women during the period he was becoming established. Although the age-specic legitimate fertility rates of Swedish women at this time were close to the maximum possible, the effective duration of the fertile period was reduced. In Ireland, there has always been a tendency for late marriage. In England and Wales there have been signicant changes in the age of rst marriage during the past 50 years (Table 10.2). The effect of the proportion of women who are married on age-specic birth rates is obvious from Table 10.3.

Factors affecting fertility in communities

It has been shown that reductions in mortality have been achieved either as by-products of tangible and universally acceptable improvements in the environment or from certain specic medical measures, such as vaccination, which reduce the risk of contracting diseases. By contrast, reductions in the fertility of a population require the consent and cooperation of individuals together with changes in their personal attitudes to child bearing and child rearing. In pretransitional populations, it is necessary for women to bear large numbers of children in order that there will be sufcient survivors to maintain the size of the community and to provide a work force to sustain essential activities. Strong social customs and belief systems have grown up to support this need and therefore changes in fertility depend on changes in social customs and ethics. Next, it is necessary to promote and make available fertility control to individuals.

Social factors
By convention, child bearing and child rearing outside marriage have been discouraged in most societies. In contemporary western societies, this attitude has changed but in many parts of the world, powerful taboos remain and societies continue to censure the unmarried mother and her child. Thus, marriage practices have a potent effect on the reproductive behaviour of societies. The legal minimum age of marriage is of less importance in most societies than the conventional

Table 10.2 First marriage rates per 1000 single women in England and Wales. (Source: Registrar Generals Annual Statistical Reviews.) Age in years 1619 1938 1948 1958 1968 1978 1988 28.1 49.1 75.2 84.6 58.8 23.0 2024 171.6 212.5 260.8 260.9 177.9 101.6 2529 132.2 158.1 162.5 161.4 134.8 106.8

Table 10.3 Births to women aged 2024 years in England and Wales. (Source: Registrar Generals Annual Statistical Reviews.) Legitimate births per 1000 married women 1939 1969 1988 252 251 212 Percentage of women who were married 33 58 30 Births per 1000 women (married and single) 93 157 95

76

Medical demography Chapter 10 The legitimate birth rate to women aged 2024 was similar in 1939 and 1969 but the age-specic birth rates differ considerably because the proportion of women who were married changed. Divorce, separation and widowhood have the reverse effect on birth rates to those of marriage. The same conventions that discourage never-married women from having children discourage divorced and widowed women from reproducing. In normal times, this has little impact on birth rates but, after the First World War, when many women in Europe were widowed, there was a noticeable reduction in the number of births, although there had been little change in the size of the female population in the reproductive age group. Sexual behaviour within marriage varies between societies. Although taboos exist regarding the permissibility of intercourse at certain times, for example during menstruation or religious feasts, this has little measurable effect on birth rates. contraception. Within Christian cultures, the Hutterite and Amish communities take the Biblical dictum to go forth and multiply quite literally and amongst them it has not been unusual for married women to produce a dozen or more children. Some non-Christian religious groups also eschew contraception on principle. Local ethics and morals may restrict the availability of the more efcient methods to certain groups. Thus, if sexual intercourse outside marriage is deemed wrong, contraception for the unmarried may be seen as a collusion with immorality. In the 1960s and 1970s, many clinics in England and some general practitioners would not advise unmarried women on contraception. In societies where the role of women is seen mainly as child bearing and child rearing, women who limit their fertility may be rejected or may fear rejection. Similar problems affect the acceptability of contraception in groups where a mans success and strength is measured by the number of children he fathers. During transition between high and low mortality, fear of death of existing infants and children, resulting in the extinction of the family, leads to the production of more children. It is often difcult to convince parents in such societies that the survival of existing children is threatened by further enlargement of the family. Even if the idea of birth control is acceptable to an individual, the method of contraception involved may be unacceptable. Many of the simpler methods require action by the male (e.g. the sheath or coitus interruptus), and they may detract from his satisfaction. The methods that require no action at the time of intercourse usually require intervention by trained professionals (e.g. the intrauterine device (IUD) or sterilization). The choice and use of methods of contraception is also affected by the couples level of education. This is important in communities where birth control is new and where modern techniques are not common knowledge. Most developing countries have recognized this factor as important and are experimenting with teaching methods. The most effective methods are usually the most expensive. If family economics mean that people cannot afford the new technology then in practice the method is 77

Contraception
Although the possibility of contraception and knowledge of techniques has existed for many years (it was known to and used by the ancient Egyptians), its use varies substantially from place to place depending on its acceptability, availability and efciency. Nowadays, in many societies, the most important social factor determining the patterns of reproduction is the acceptability of contraception. In general, the better educated (and those who are better off) are more likely to use contraception than the ill-educated and poor. Its use is also determined to some extent by religious beliefs. Members of the Roman Catholic Church are forbidden to use articial methods of birth control. Nevertheless, the rule of the church is not universally adhered to and contraceptive practice varies amongst Roman Catholics. It has been shown that a large proportion of Roman Catholics in Europe and North America no longer adhere to their churchs teaching. The Roman Catholic Church is not the only religious group actively to discourage the practice of

Chapter 10 Medical demography not available to them. The problem of cost is greatest in countries with the greatest problems. The efciency of a particular method of contraception is assessed by the number of conceptions per women-years of use. The assessment should be made in a group similar to that in which the method will be used. Table 10.4 shows estimates of relative efciency of some of the current methods. These estimates were made in married women who were likely to have regular intercourse and to be motivated to use the method correctly. Some couples use contraceptive methods incorrectly. For example, there is some evidence that some women are erratic in their use of oral contraceptives which alters the apparent effectiveness of the method. In 1969 and again in 1995, a great deal of publicity was given to the risk of venous thromboembolism associated with oral contraceptives and a large number of women precipitately stopped using them. They did not appear to use alternative methods and consequently the decline in birth rate in England and Wales was temporarily halted (Fig. 10.10).
Table 10.4 The relative efciency of different methods of contraception. Pregnancies per 100 women-years of use 0.15 2.00 2.405.00

Some recent changes in the patterns of fertility in England and Wales

For about 50 years until the 1980s, there was a tendency for women to marry earlier. Since then there has been a steady rise in the age of rst marriage. The mean interval between rst marriage and the birth of the rst child fell until the early 1970s when it began to increase (Fig. 10.11). The increase in the interval between rst marriage and rst pregnancy was associated with an increase in the use of efcient contraception, particularly oral contraception and the IUD. The mean interval

94 90 86 82 78 74 70

General fertility rate

28 24 20 16 12

Oral contraceptive sales 1964 65 66 67 68 69 70 71 72 73 Year

8 2

Contraceptive used Oral contraceptives Intrauterine device Diaphragm

Figure 10.10 General fertility rate and oral contraceptive sales, showing the effect of the 1969 pill scare.

30 25 20 15 Interval in months 10 5 Age at first birth Age at marriage

30 25 20 15 10 Figure 10.11 Average age of women at marriage and average age at birth of rst legitimate child in England and Wales.

1938 43

48

53

58

63

68

73

78

83

Year (193883)

78

Months

Years

Packets of oral contraceptives (1000 000s)

Total births per 1000 women aged 1544

Medical demography Chapter 10 between marriage and pregnancy is affected by the proportion of women who are pregnant when they marry. Figure 10.12 shows that in the 1960s about 40% of women who married under the age of 20 years and about 15% of women aged 2029 years were pregnant when they married. The proportions fell in all age groups during the 1970s. The post-1970s changes were due to a combination of increased availability of abortion and of contraception to unmarried people. This hypothesis is consistent with the fall both in the illegitimate birth rate and in the number of marriages of pregnant women. Figure 10.13 shows the cumulative age-specic fertility rates for cohorts of women born in different years. The 1920 and 1930 cohorts reached their peak birth rates at about the age of 26 years and fertility was high well into the 30s. By contrast, the 1940 cohort reached its peak fertility at age 24 years and tended to have more children earlier in their lives. The 1950 cohorts fertility was stable between the ages of 21 and 28 years. It is probable that family size of the pre-1941 cohorts was determined largely by the age of marriage and that, within marriage, conscious control of fertility was haphazard, whilst the post-1941 cohorts married earlier and exercised a more precise conscious control over fertility.

Total-period fertility rate

This is a useful measure calculated from summing the age-specic fertility rates and expressing the sum of the rates as the expected number of live births per woman of child-bearing age. Thus, in Table 10.1, we can see the UK total-period fertility rate is 1.84 which is below the replacement level of 2.0. In fact, because some children die before they reach reproductive age, the replacement totalperiod fertility rate is about 2.1 in the UK, and in countries with high infant and child mortality the rate will be even greater.

40 Marriages with a birth within 8 months (percentages) 35 30 25 20 15 10 5 0 1952 57 62 67 72 77 Year (195287) 82 87 Age 2529 Age 2024 Age 1619

Fetal loss and infant mortality

Fetal and infant survival rates are amongst the most important factors inuencing demographic change. Fetal loss during pregnancy occurs in three ways.

Figure 10.12 Trends in known illegitimate conceptions in England and Wales.

2500 Births per 1000 women 2000 1500 1000 500 0 15 20 25 Age Figure 10.13 Cumulative fertility of birth cohorts in England and Wales. 30 Years women born 1920 1930 1940 1950 35 40

Fetal loss
Spontaneous abortion Induced abortion Stillbirths

In developed countries, 1525% of known conceptions spontaneously abort. The true rate may be as high as 40%. Sixty per cent of spontaneous abortions have abnormal chromosomes. In the process of demographic transition, changes in sponta79

Chapter 10 Medical demography

160 140 Total known conceptions (1000s) 120 100 80 60 40 20 0 1969 1974 Year 1979 1984 Figure 10.14 Conceptions in women under 20 years, distinguishing those leading to maternities from those terminated by abortion. Terminations Maternities

neous abortion and stillbirth rates are not signicant elements. Induced abortion depends upon individual motivation and it affects age-specic birth rates selectively. In countries where induced abortion is legal, full statistics are published. Figure 10.14 shows the numbers of known conceptions in women aged 1619 years in England and Wales from 1969 to 2001, and demonstrates the contribution of legal abortion to the fall in birth rate. Perinatal and infant mortality rates are sometimes used as sensitive indicators of the quality of health services within a country or within a district. This is asserted because some of the causes of perinatal and infant deaths are avoidable by medical intervention. Three major studies of perinatal deaths in Britain (1946, 1958 and 1970) involved following up cohort of births beyond the perinatal period to examine factors related to perinatal mortality and morbidity. They showed that adverse maternal obstetric factors act in a cumulative manner. Perinatal mortality rates (PMRs) are highest in para 3 women, in women at the end of their reproductive life and when the birth interval is less than 12 months or more than 60 months. Conversely, they are lowest in para 1 women, in women aged 2029 years and when the birth interval is 1835 months. The PMR is higher for illegitimate births than it is 80

for legitimate births, even after account is taken of parity and maternal age. There is a positive social class gradient, i.e. social class V has PMRs greater than social class I. Some social class differences are due to reproductive behaviour. Birth weight is highly correlated with perinatal mortality. The proportion of low birth weight babies born within a country is a strong determinant of its PMR. Also, there is a close correlation between low birth weight and certain maternal factors, for example parity, birth interval and maternal age. Poor maternal health can also adversely affect PMRs. Important diseases or conditions that have been shown to be associated with high PMR include the following.

Factors associated with high PMR

Hypertension Poorly controlled diabetes Renal disease (which can also decrease fertility) Infection (hepatitis B, syphilis, rubella, cytomegalovirus and toxoplasmosis can cause fetal abnormalities) Severe malnutrition Smoking Alcohol can cause fetal alcohol syndrome (intrauterine growth retardation, developmental delay and spontaneous abortion)

Medical demography Chapter 10 Thus, while a large proportion of fetal and perinatal mortality is difcult to prevent, much can be done to reduce rates by appropriate antenatal and postnatal care and advice. reduction in mortality in developing countries are malnutrition, illiteracy and poverty. Industrialization is inversely related to changes in fertility. Four explanations for this can be adduced as follows: in urban societies children are not an economic asset; as the infant death rate declines, the proportion of children who survive to adulthood increases and the number of births required to attain a desired family size is smaller; in urban societies, there are greater opportunities for women outside the domestic environment, and being committed to child rearing restricts a womans activities; and in educated societies, the inuence of secular rationality is stronger which allows readier acceptance of contraception.

Summary
Every industrialized nation has low mortality compared with non-industrialized countries. Further substantial decline in mortality in industrialized countries is unlikely because the major causes of death are associated with old age. There is great potential for further substantial reduction in mortality in Asia, Africa and Latin America. This will be achieved by control of the major infective diseases, especially gastrointestinal and respiratory infections in children and AIDS. The principal factors acting against any quick

81

Chapter 11 Evidence-based medicine

Introduction
One of the guiding principles behind the new NHS is that health care should be based on evidence. The idea of evidence-based practice was introduced into the UK relatively recently. The rst workshop was organized by Professor David Sackett at Oxford in 1995. Medicine has not always been evidence based that is why evidence-based medicine (EBM) has been regarded as a new discipline. Traditionally the teaching of medicine has been based on an apprenticeship-type system. Undergraduate students observed their teachers as they practised, learnt from them and emulated them. This same tradition was continued in postgraduate training. EBM is the application of population-based research to the care of patients. Increasingly patients and those responsible for paying for health care expect evidence that the treatments offered are of proven benet. They want to know if a drug is prescribed that it not only lowers the cholesterol level or reduces the blood pressure but that it also prolongs life or reduces the risk of heart attack or stroke. Many illnesses or diseases have a range of treatments, all of which are effective. Some treatments are of unproven value and others although effective can cause harm. For other conditions there are no effective treatment. When evaluating interventions in an environment of limited resources it is desirable to compare the relative bene82

ts and risks of available treatments (including no treatment). Benets and harm can be assessed in terms of either cost, or some measure of health outcome. Much medical practice is based on anecdotal evidence and experience which may be unreliable and biased. EBM promotes the concept that clinical (and health care) decision-making should be based on the best patient/populationbased studies. It uses a hierarchy of evidence, with the highest quality normally coming from randomized controlled trials (RCTs). These are not always possible and in such circumstances evidence from other studies is used.

Hierarchy of evidence
Systematic review of randomized trials Randomized controlled trial Cohort studies Casecontrol studies Case series or case reports

EBM came about because of the revolution in information technology. The rapid growth in the availability of electronic databases of the medical literature allowed original research papers to be identied and retrieved rapidly. This can be done without leaving the consulting room. Before the advent of computerized databases many doctors

Evidence-based medicine Chapter 11 relied on medical textbooks, supplemented by browsing the journals that crossed their door or picking up new ideas from observing the practice of consultant colleagues. Unfortunately the textbooks are usually out of date and many of the journals were little more than medical newspapers, which published articles that were newsworthy rather than of scientic merit. Consultant practice was followed because it was believed that consultants were better informed of new developments and better equipped to appraise new studies critically. Whilst this was often true, there can be much variation in consultant practice and this can lead to uncertainty in decision-making. structure their questions. (P is for population, I for intervention, C for comparison and O for outcome.) PHow are a group of patients similar to the one in the clinical scenario described? Patients included in the published studies may differ in some respects to the patient or clinical scenario encountered. Key factors such as age, gender, diagnosis, ethnicity and so forth need to be considered. IWhich main intervention, treatment (or exposure or prognostic marker) is being considered? CWhat is the main alternative with which this manoeuvre or treatment should be compared? Many studies on new therapies are compared against placebo. This is required by the licensing authorities to demonstrate efcacy but clinicians want to know whether the new treatment is more effective than the treatment usually offered. OWhat can be accomplished with the patient e.g. increased length of life, shorter hospital stay, less pain, etc? The outcome must be measurable. There are objective measures of health or disease that can be used to assess one treatment or group against another.

Practising EBM
EBM is used by clinicians to help their decisionmaking. It utilizes a structured approach, involving ve key steps. These include taking a clinical scenario and from this identifying the key questions that are needed for the management of the patient. That question should then be formulated in such a way that it can be answered through use of the medical literature. The structured question produces key words which are used to help formulate the search strategy that identies the relevant papers. The papers then have to be critically appraised and the evidence synthesized and used to help clinical decision-making.

Searching for the evidence

The key search terms are found from the structure of the question, particularly the intervention/exposure of interest and the outcome of interest. The search strategy should be as sensitive as possible it should identify all the relevant papers to help answer the clinical question. Specicity is obtained by combining search terms and by using the relevant lters for the different types of question. (For example, a lter can be developed that restricts the search to randomized trials.)

Structured approach to EBM

Clinical scenario Structured question Search for the relevant papers Critically appraise the evidence Use evidence to help decision-making

Critically appraising the evidence Forming structured questions

Forming an answerable question from a clinical scenario is the rst step in practising EBM. It is a discipline that requires practice. Practitioners of EBM often use the acronym PICO to help them Once a relevant paper has been identied it is important to be able to appraise it critically. This is done by asking three simple questions. How valid is the study? Is it well designed and carried out in an appropriate population? 83

Chapter 11 Evidence-based medicine What are the results of the study. This often entails unpicking the results and presenting them in a way that is more relevant to the question. Will the results of this study affect practice, i.e. is it clinically relevant? It is important to be aware that a statistically signicant nding is not necessarily clinically relevant. noted that if 20 characteristics are looked at, then by chance (at the 5% level) a signicant difference between the groups is likely to be found in at least one. The larger the study the more likely the groups are to be similar. Thus big studies are to be preferred. This will also help avoid the problem of a Type 1 or a Type 2 error (see below).

Drop-out rates

Randomized controlled trials

The principal form of evidence when considering whether a treatment works or whether an exposure causes a particular outcome is an RCT. Hence it is important to understand the principal components of an RCT. The study should be appraised critically to see whether it has been well conducted and can be believed.

Points to look for in appraising an RCT

Randomization Characterization of the groups Drop-out rates Intention to treat analysis Blinding Sample size Results

Were all patients who entered the trial properly accounted for and attributed at its conclusion? If patients are lost to follow-up it may be that those patients who left the study had a different outcome to those who were included in the nal analysis. For example, if the outcome of interest is death, patients lost to follow-up may have had a higher death rate than those who are followed up throughout the study. This leads to an underestimate of mortality in the groups studied. Similarly, if the drop-out rate between groups is different bias may be introduced. A common reason for a difference in drop-out rates is that one treatment causes more side-effects or is ineffective. Ideally all patients should receive the treatment to which they were randomized, be followed up and their outcomes noted.

Intention to treat analysis Randomization

The method of randomization can introduce bias and inuence the generalizability of the ndings. It is therefore important to know how the randomization was carried out in order to be able to assess whether this is likely to have inuenced the results. It is particularly important to check that the staff involved in recruiting subjects to the study were not also responsible for the randomization. In reality there are always some patients who do not receive the treatment to which they were randomized. By analysing the results using intention to treat analysis any bias due to unplanned drop-outs or cross-overs will be avoided. (A cross-over is when a patient is allocated to treatment A but actually receives treatment B.) If on intention to treat analysis there is still a treatment effect then this is likely to be a true effect. Analysis by actual treatment group is also usually worthwhile and is more likely to show a statistically signicant difference between groups. With placebo-controlled trials it has been shown that compliant patients who take their placebo have a better outcome (up to 30% better) than the non-compliant patients. Including dropouts in the placebo group can introduce bias. Again, intention to treat analysis removes this bias.

Characterization of the groups

In the paper there should be a table showing the characteristics of the treatment groups being compared. Sometimes by chance, particularly in small studies, the groups may be unequal (e.g. more men in one group) and this can cause bias. It should be 84

Evidence-based medicine Chapter 11

Blinding
In a single blind randomized trial the patient is unaware which treatment they are receiving. This is important when the assessment of the outcome of interest is subjective such as pain, anxiety, etc. In a drug trial this can be achieved by giving one group of patients a placebo. In a double-blind trial both the patient and the investigator should not know which treatment they received. This then removes possible bias in both the reporting and recording of the outcome of interest.

An estimate of the response rate in one of the groups if the outcome of interest is a rare event, then a larger sample size will be needed. Level of statistical signicance this is usually at least the 95% level to avoid a Type 1 error. The value of the power desired this is usually set at the 80% level because missing a small but true difference is less important than identifying a spurious positive association. Greater power requires a bigger sample size and will incur additional costs.

Sample size
Small studies can sometimes be misleading. A Type 1 error occurs when a study concludes that two treatments are different when in fact they are not. If a study is repeated 20 times, on one occasion by chance a statistical difference (at the 5% level) will be detected. This difference occurs by chance but if the results of only this one study are published it will give a biased impression that the treatment investigated is worthwhile when in fact it is not. A Type 2 error is when the study concludes that the treatment groups are not different when in fact they are. In this case, unless a big enough study has been carried out the difference will not be detected a Type 2 error. It may be that although there is a true difference between the two interventions, the size of this difference is small and may not be clinically relevant. Thus a Type 2 error is often not considered such a serious problem as a Type 1 error.

The results
The results of an RCT are usually presented as a relative risk (see p. 11). Relative risk (RR) is the absolute risk in the treated group divided by the absolute risk in the untreated group (or vice versa). Another way the results can be presented is by estimating the number needed to treat (NNT). This takes into account not only the RR but also the absolute risk in the two groups being investigated. The NNT is calculated by taking the reciprocal of the absolute risk reduction (ARR). It indicates how many people would have to be treated with A as compared to B in order to prevent one additional outcome of interest. For example, imagine 2000 patients with mild hypertension are randomly allocated to treatment or placebo. At the end of the year 4 patients in the placebo group have had a stroke and only 2 in the treated group have suffered a stroke. The RR for the treated group if 0.5. Thus the treatment produces a 50% reduction in the number of strokes. However the NNT in this example is 500. Five hundred people will have to be treated for 1 year (and carry any risks associated with that treatment) for one patient to benet. NNTs are probably a more relevant index to be used for clinical practice although there is evidence that decision makers are more likely to alter their practice when presented with the RR.

Groups are not different Conclude groups are not different Conclude groups are different Correct decision Type 1 error

Groups are different Type 2 error Correct decision

When designing a study it is important to ensure that Type 1 and Type 2 error is avoided. The size of the sample required is determined by: Difference in response rates to be determined this should be a clinically signicant difference.

Critical appraisal of systematic reviews

A systematic review involves identication of all the relevant primary papers in human populations 85

Chapter 11 Evidence-based medicine that deal with a focused question. These papers are then appraised critically to identify their strengths and weaknesses. Finally a summary of the evidence is reported. Well done systematic reviews are increasingly being accepted as the highest form of evidence in the hierarchy of evidence. The gold standard for reviews are those done to the criteria set down by the Cochrane Collaboration a group of researchers and clinicians interested in undertaking systematic reviews of randomized trials. Since a systematic review is a retrospective look at published papers, it is important to make the process rigorous and well dened to prevent bias and thus distortion of the ndings. searched, e.g. Medline, Cochrane, Embase, Cinahl, etc. Details of secondary references, i.e. the references cited by the papers that were retrieved from the original search. Studies published in languages other than English. Researchers undertaking systematic reviews are often tempted to exclude these papers because of the cost of having them translated (as well as the delay). However this may mean that some perfectly valid studies that deal with the question are not included. This again can introduce bias. Grey literature. Studies with negative ndings may be difcult to get published. If only studies with positive results are published then the published papers will give a positive result. The results of unpublished studies (the grey literature) if relevant should be included. They can be obtained by contacting researchers known to be active in the eld of interest. Also drug companies often have unpublished studies which they may release to researchers. Once all the papers have been collected then they should be appraised critically and an evaluation of the overall ndings made.

Points to look for in appraising a systematic review

Criteria for inclusion Sensitivity of the search Method of selection Validity of the studies

Criteria for inclusion

A systematic review should have clearly dened criteria for the inclusion of studies. This usually includes the type of study (for therapy questions ideally an RCT), the populations included in the studies, the treatments or exposures and relevant outcomes. The criteria should not be so restrictive that important studies are likely to be missed. Ideally the authors should list all trials reviewed with a reject log and reasons for exclusion.

Meta-analysis
A meta-analysis is a particular type of systematic review that uses quantitative methods to combine the results of several independent studies considered by the analyst to be combinable. The overall results are weighted by the size of the contributing studies. This means that the larger studies will have the main inuence on the outcome. The results of a meta-analysis can be presented in a tabular or graphical form.

Sensitivity of the search

The systematic review should demonstrate that a sensitive search strategy was adopted. Ideally it will include: The words that were used in the interrogation of the medical databases, which terms were combined and which intersected. The time period over which papers could be included. The databases and other sources that have been 86

Challenges to evidence-based practice

There is a word of warning about the rational or scientic approach to medicine. Firstly there is a mismatch between the needs of patients and the research agenda that provides the evidence. Research agendas are set by those with the funds particularly the research councils and the

Evidence-based medicine Chapter 11 pharmaceutical industry. The Medical Research Council has been criticized for the lack of involvement of patients in setting research agendas. Rather, it is a panel of scientists with a particular view of what research is needed that commissions studies. Their viewpoint has a major inuence on the type of research that is funded. The pharmaceutical industry, which provides more than 60% of the funds for medical research in the UK, naturally has a different perspective. They want to nd medications that will alleviate patients problems and will lead to commercial success. Investigations of behavioural or population-based interventions have a much harder time attracting funds. Consequently there is more evidence on the effectiveness of interventions that involve drug treatment than other modalities. Studies on new drugs often exclude specic groups for ethical or safety reasons. For example, there are few studies amongst women of childbearing age, children and the elderly. Practising EBM can be difcult for obstetricians, paediatricians and geriatricians. A consequence of excluding certain groups from RCTs is well illustrated by the recommendations for the use of lipid-lowering therapy. It has been suggested that only those who are at high risk of heart disease and who are under 75 years of age should be treated. Part of the reason for the age cut-off is that the trials of lipid-lowering therapy excluded older patients (above 70 or 75 years of age). This was not because these patients are unlikely to benet but because the likelihood of side-effects and adverse events are higher in older people. Consequently it makes sense to try to avoid including these patients in the trials. Unfortunately there is then no direct evidence from the RCTs of the outcomes in older patients. The second problem is that EBM often clashes with clinical experience and does not take account of the context in which practice takes place. It has been shown that the occurrence of an adverse event when treating a patient was one of the biggest barriers to following treatment suggested by published studies. Doctors have always had the clinical freedom to make judgements about the best course for individual patients, balancing their knowledge of the patient, their clinical experience and the evidence from the literature. Increasingly with the publication and dissemination of protocols and service frameworks the ability to balance experience with evidence-based practice is being eroded. Thirdly there is the issue of the balance between the patients experience and understanding and that of the health service. Clinicians are encouraged to respect the autonomy of patients and to take into account their views and experience when offering them treatment. But often these conict with evidence from the published literature. Which should take precedence?

Strengths and weaknesses of EBM

EBM has a number of strengths and weaknesses. Its practice requires basic skills in searching databases of the medical literature, skills in epidemiology to help appraise the relevant papers and skills in statistics to help interpret the results.

Strengths
Helps clinicians in their decision-making Helps ensure consistency of care offered to patients Develops skills in critical appraisal Helps clinicians keep up to date Helps in the development of evidence-based guidelines

Weaknesses
Development of the evidence base has been biased Ignores the benet of clinical experience Does not take into account patient choice Can be time consuming and requires the acquisition of basic skills

87

Part 2 Prevention and Control of Disease

Chapter 12 General principles

Introduction
The health of a population depends both on the provision of health care for the sick and on public health services to promote health and prevent the spread of disease. Until the middle of the 19th century the state accepted little responsibility for health. The rst attempts to improve public health in the UK involved legislation, beginning with the Nuisances Act of 1846, which aimed to remove sewage and offal from the streets. This was followed by a series of further public health acts. At the beginning of the 20th century the Government turned its attention to personal medical care. David Lloyd George introduced insurance-based health care for workers in 1911. The majority of health care, however, was still obtained privately, or through friendly societies and charitable institutions. The state took no major role as a provider of health care until the inception of the NHS in 1948, which promised access to free health services for all. Today, the Secretary of State for Health is responsible to Parliament for the work of the Department of Health (DoH) whose aim is to improve the health and well-being of people in England (see DoH website at http://www.doh.gov.uk/). Separate arrangements are in place for Scotland, Wales and Northern Ireland. Earlier generations tended to accept ill health and premature death as unavoidable hazards of human existence. Over time people have come to

expect a long and healthy life. Nowadays if illness occurs it is assumed that modern medicine can or ought to be able to restore the sufferer to normal health. These changed expectations have been brought about to a large extent by the publicity given to the more dramatic advances in medical knowledge and treatments and by the evident success of modern medicines in reducing mortality, particularly during infancy and childhood. The public also feels a sense of ownership of the health service and expects ready access to it when needed. Although it is true that during the past 50 years the scope and effectiveness of medical treatments have been extended greatly, it is also true that many of the diseases which commonly affect humans are self-limiting and that medical treatment does little to alter their natural course. Furthermore, few of the diseases that result in death or major disability can be cured. The main impact of modern medicine has tended to be to allow people to live longer and more comfortably with their diseases. The public often fails to appreciate this. For many of the major diseases, it is both logical and desirable to take steps where possible to prevent their occurrence. Even if a treatment eventually becomes available, a strategy of prevention would usually be more cost effective in improving both public and personal health. In future, it is hoped that medical research and practice will give greater attention to the means whereby health can be promoted and diseases prevented. Historically, infec91

Chapter 12 General principles tious diseases were the major causes of morbidity and mortality, particularly in children and young adults. Their control over the past 150 years owes more to social and economic progress than it does to specic medical intervention. Preventive programmes during this period have included such measures as improvements in sanitation, water supply, the quantity and quality of food and the quality of housing, safer conditions in the workplace and raised standards of personal hygiene. All of these carry obvious and immediate benets other than those purely related to health: they make life more comfortable and pleasant with little or no restriction on personal freedom. Most of the changes were at community level and were the result of legislation rather than action by individuals. This made them comparatively easy to institute. By contrast, some of the more recent advances in the control and prevention of communicable diseases, such as the elimination of diphtheria and poliomyelitis in many countries and the worldwide eradication of smallpox, required mainly medical action (immunization) and thus can rightly be claimed as major medical achievements. The benets of environmental improvements, as well as of specic immunization, however, will be sustained only by continued vigilance. Much modern preventive medicine is directed to this end. In the past, the presence of a disease in the community served as a constant reminder of its nature and consequences. In societies dependent upon distant memories of childhood infections such as measles, whooping cough, polio and tuberculosis, continuing public education is essential to sustain preventive activities. With the exception of smallpox the causal organisms have not been eradicated from human populations. Thus the diseases can recur. The virtual elimination of the older lifethreatening infectious diseases has brought the non-infectious illnesses into greater prominence. In modern times, despite the emergence of new infectious disease such as legionnaires disease, HIV and severe acute respiratory syndrome (SARS), it is cardiovascular disease, malignancies, degenerative conditions (such as arthritis) and other chronic illnesses which occur amongst older people that are the major health problems. The prevention of many 92 of these diseases is more complicated than the control of infectious diseases and therefore progress is more difcult to achieve. There has been a proportional increase in accidents as a cause of morbidity. The problems of prevention of chronic diseases centre around their natural history, the difculty in identifying aetiological agents and the fact that many have multiple causes. Moreover, they are generally characterized by having a long latent period between exposure to the aetiological agent and the appearance of symptoms. In many cases, the symptoms have an insidious onset and by the time they are of sufcient severity to cause the affected individual to seek medical attention, irreparable damage has been done. Prevention of these diseases often depends on actions by the individual, rather than passively enjoying improvements in the environment brought about by the actions of others. It demands modication of personal behaviour in such matters as the use of tobacco and alcohol, diet and exercise at a time in life when the risks of contracting the disease in question are seen as remote. It is also a fact that, even for common diseases, the absolute risks for the individual are indeed relatively small. In these circumstances, campaigns to persuade people to change their lifestyle require great skill and patience sustained over long periods of time. These lifestyle changes also need to be complemented by public policies that promote health by, for example, the taxation of tobacco and alcohol products, the subsidizing of food production and the provision of public recreational facilities. These all require a political will to be implemented. Despite the difculties, prevention remains an important aspiration and progress is being made in some of these diseases (e.g. in reduction of cancer mortality), both by action at a political level and by persuading people to change their lifestyle and habits. The interaction between the social and physical environment and health has also been much more widely recognized in the last 30 years by national and international bodies such as the World Health Organization (WHO). It has led to the concept of the promotion of a healthy environment and lifestyle being adopted in a number of cities. Acknowledgement that employment, housing, bal-

General principles Chapter 12 anced diets and a social and economic environment that promotes health are all important in improving the quality of peoples lives and increasing the length of life has meant that both government and local policies which affect social factors have to take into account the long-term consequences to health. progress is rapidly arrested or reversed or its consequences minimized. Useful preventive action does not necessarily require knowledge of the cause of a disease. There are many examples of effective prevention that preceded discovery of the agent or complete understanding of the causal mechanism. For example, in the 18th century, Lind (Fig. 12.2) and Blane demonstrated that scurvy in the crews of ships could be prevented by the consumption of adequate amounts of citrus fruit; this was long before vitamin C was discovered. In the 19th century John Snow (Fig. 12.3) showed that cholera was transmitted by drinking water polluted by sewage. His ndings led to the elimination of cholera by the provision of pure water supplies many decades before the isolation of the causal organism. In this century, Doll et al. (see Chapter 5) demonstrated that those who stop smoking cigarettes substantially reduce their risk of contracting lung cancer, though the carcinogenic agent in tobacco smoke has yet to be identied. In general, however, a full and accurate understanding of the causes

Principles of prevention
Disease is the result of a harmful interaction between the host (humans), a pathogenic agent and the environment (Fig. 12.1). Agent, host and environment form a dynamic system in which, in the healthy individual, the balance normally favours the host. Thus, if the agent is locally absent or contained, or its capacity to cause disease is matched by the hosts protective mechanisms, or the environment inhibits the spread of the agent, health is maintained. Disease or injury occurs when the balance is disturbed, for example owing to changes in the pathogenicity of an agent, changes in environmental conditions that favour the survival and transmission of the agent to humans, or the breakdown or absence of human normal defence mechanisms. The control and prevention of disease depends on effective intervention in the relationship between agent, host and environment to ensure that the balance remains in the humans favour, or, if disease does occur, to ensure that its

Affects presence and survival of agents

Environmental conditions: Physical Biological Social Affects exposure of humans to agents

Affects human capacity to resist diseases

Agent properties: Microbial Chemical Physical Psychological

Human protective mechanism: Immunity natural (non-specific) acquired (specific) Behaviour

Figure 12.1 Interactions of agent, host and environment, causing disease.

Figure 12.2 James Lind (171694) author of the treatise on scurvy.

93

Chapter 12 General principles

Strategies related to the agent

If the agent can be identied, it may be possible to remove or destroy it at source. For example, by ceasing to use asbestos as an insulating material, the incidence of mesothelioma has been reduced; the control of bovine tuberculosis in humans was achieved by eradication of the disease from milking herds and pasteurization of milk.

Strategies related to the environment

These include attention to general environmental factors such as standards of housing, nutrition, working conditions, water supplies, sewage disposal and the control of environmental pollution. Environmental measures directed at the specic causes of individual diseases are also important and people may be protected from potentially injurious agents by the construction of barriers between them and the source of harm. Examples of such measures include the prevention of transmission of food-borne infection by hygienic food production methods; elimination of vectors, for example action to prevent the spread of malaria or yellow fever by mosquito control; and the use of machine guards in industry to reduce the risk of accidents.

Figure 12.3 John Snow (181358) epidemiologist who studied the transmission of cholera.

of diseases and of the factors that determine the balance between agent, host and environment is helpful in order to construct appropriately directed preventive and control programmes. Epidemiological studies are used to identify the causal agents and those elements in the environment or in peoples behaviour and personal characteristics that are key determinants of the natural history of disease.

Strategies related to humans

There are three strategies involving individuals. The enhancement of general or specic resistance to disease, i.e. by improved nutrition or immunization. The modication of personal behaviour, i.e. by encouraging people to adopt healthier lifestyles by not smoking, moderating alcohol intake, improving diet, avoiding obesity, exercising regularly, etc. The use of screening to detect predisposing conditions or the early stages of disease when action can be taken to prevent its onset or control its progress, for example tuberculin testing for tuberculosis, blood pressure measurement to identify hypertension, or mammography for breast cancer detection.

Intervention strategies
Based on the knowledge gained from epidemiological studies three main types of intervention strategy may be adopted.

Intervention strategies
Strategies related to: Agent Environment Humans

94

General principles Chapter 12 ple with manifest disease by modifying continuing risk factors such as smoking and by the implementation of effective rehabilitation.

Preventive action
Action is usually classied as follows. Action
Primary prevention: prevents disease starting Secondary prevention: detects disease early Tertiary prevention: damage limitation

High-risk individual vs. population strategy

Where a choice of strategy exists, the planning of a preventive programme should take account of certain practical considerations. The most desirable approach is one that gives the greatest benet to the largest number of people. In some instances, this may mean that the most effective strategy is to target high-risk individuals. Such programmes, whilst of benet to individuals, may do little to reduce the overall burden of disease in a population. Sometimes a population-based approach which confers a smaller benet on a large number of individuals may yield greater dividends. The population strategy has the advantage that there is no need to identify a high-risk group. Everyone is targeted. Interventions that are simple and require minimal cooperation from individuals are usually the most successful. Economic factors must also be considered when deciding on the most appropriate intervention strategy. Each of these strategies for prevention is considered in detail in the chapters that follow.

Primary prevention
This aims to prevent a disease process from starting. It often calls for strategies directed at the removal or destruction of agents but can also include environmental control, immunization, health promotion and health education.

Secondary prevention
This aims to detect disease at the earliest possible stage and to institute measures to cure or prevent its further progression. Screening programmes backed by effective interventions are the most important examples of secondary prevention.

Tertiary prevention
This is concerned with damage limitation in peo-

95

Chapter 13 Health promotion and health education

Introduction
The terms health promotion and health education are sometimes confused. Both are strategies aimed at improving the public health, but while the concepts are complementary they are not synonymous. Health promotion involves the empowerment of the community in improving its health through education, through the provision of preventive health services and by improvement of the social, physical and economic environments. Health education is the empowerment of individuals through increased knowledge and understanding, but does not involve the political advocacy necessary in health promotion. The health strategies that emerged during the 19th century were in some ways similar to those that we now term health promotion. Thus, Medical Ofcers of Health worked for local authorities with the aim of improving the environment, encouraging healthy public policies, introducing preventive strategies (e.g. sanitation and vaccination) and encouraging better health through education. Another step in the development of health promotion was the Peckham Pioneer Health Centre project, which began in south London in the 1930s. It provided conventional health care and health education together within an environment that supported community development through the provision of recreational and sports facilities. 96

The new public health

A new public health initiative was heralded by the Lalonde Report for the Canadian Government (1974), which incorporated health promotion as an integral part of the government strategy to improve public health. Lalonde identied four main inuences on peoples health.

Lalondes four health factors

1 Genetic and biological factors 2 Behavioural and attitudinal factorsthe so-called lifestyle factors 3 Environmental factors, which include economic, social, cultural and physical factors 4 The organization of health care systems

A growing awareness of the factors that inuence health encouraged people with an interest in prevention to involve organizations and institutions not usually primarily concerned with health. This led to the concept of Healthy Cities, which also originated in Canada and was subsequently embraced by the World Health Organization (WHO), spreading throughout the world. In the UK, many health promotion initiatives were coordinated under this umbrella, rst in Liverpool and later in Manchester, Newcastle, Camden, Belfast and Glasgow. More information about Healthy

Health promotion and education Chapter 13 Cities can be found at the WHO website www.who.dk/healthy-cities/. At the same time the role of the UK Health Education Council, which was set up in 1968, was expanded to include public policy advice and social and environmental issues in addition to the provision and distribution of health education material. The key components of health promotion were dened in a charter agreed at the rst International Conference on Health Promotion held in Ottawa in 1986. This suggested a denition of health promotion and ve key areas for action. The Ottawa Charter stated that: Health Promotion is the process of enabling people to increase control over, and to improve, their health. To reach a state of complete physical, mental and social well-being, an individual or group must be able to identify and to realize aspirations, to satisfy needs and to change or cope with the environment. Health is therefore seen as a resource for everyday life, not the objective of living. Health is a positive concept emphasizing social and personal resources, as well as physical capabilities. Therefore, health promotion is not just the responsibility of the health sector, but goes beyond healthy life-styles to well-being. It also proposed that: health promotion should focus on equity in health and reducing differences in health status by ensuring equal opportunities and resources to enable all people to achieve their fullest health potential. The ve areas for health promotion action were as follows. identify and remove obstacles to healthy policies so that these become the easier choice. Creating supportive environments To create living and working conditions that are safe, stimulating, satisfying and enjoyable. To encourage communities to care for each other, and to take responsibility for the conservation of natural resources. Strengthening community action To work through effective community action in setting priorities, making decisions, planning strategies and implementing them to achieve better health. Developing personal skills To support social and personal development through the provision of information, health education and the development of individual skills. Reorientating the health services To encourage health service providers to look beyond their mandate for clinical and curative services and ensure that health services are aimed at the pursuit of health rather than only the cure of illness. The principles of the Ottawa Charter were adopted in various ways by many countries throughout the world, but the initial enthusiasm seems to have waned. The UK adopted health targets in line with Health for All by the Year 2000 in 1990, and in 1999 a new set of goals were outlined in Our Healthier Nation. These targets are aimed primarily at action by the health services without a commitment to changes in public policy. They include targets to improve health outcomes in relation to cancer, coronary heart disease and stroke, accidents and mental health. There are a number of difculties in adopting the health promotion approach. The long interval between the adoption of preventive strategies and measurable improvements in health means that organizations see little short-term return on their investment. The processes of community consultation, health education and altering public policies are time consuming, and are often politically controversial. Many health promotion programmes have been initiated without a clear commitment to evaluate 97

The Ottawa Charter

1 2 3 4 5 Building healthy public policy Creating supportive environments Strengthening community action Developing personal skills Reorientating the health services

Building healthy public policy To encourage policy makers in organizations and government to place health on their agenda. This may include efforts to

Chapter 13 Health promotion and education their outcomes. Given the limited health budget, it is not acceptable to institute unproven interventions, whether they involve conventional medical treatment or a health promotion programme, unless they are rigorously and scientically tested. The emphasis that many politicians and others have placed on personal responsibility for health has been criticized because it ignores the economic and social inuences. This can be illustrated by considering smokers who suffer ill health. They are blamed for the outcome of their voluntary action whilst the advertising of tobacco products in many countries continues to be permitted and the companies who promote them take no responsibility for the adverse outcome. Similarly, children who grow up in impoverished homes, lacking education and with little hope of employment, have bleak futures and may be unable to respond to the admonition of those from more privileged backgrounds to change their ways. (These issues were discussed in the Black Report referred to on p. 5.) Another issue relating to the effectiveness of health promotion programmes concerns the dilemma of whether to adopt a population strategy or a targeted strategy. The former involves attempting to achieve health gain through actions involving the whole population while the latter focuses efforts on particular risks associated with specic conditions. Both approaches have their adherents, but scientic evaluation of their comparative effectiveness is needed before one approach or another is taken. An example of a population approach was the North Karelia Community trial, which aimed to reduce the incidence of heart disease in a Finnish community by means of changes in peoples diet, smoking habits and exercise compared with a control community. Health promotion campaigns targeted at particular groups have also been used successfully, for example in the effort to reduce the spread of HIV amongst intravenous drug users by the introduction of needle-exchange schemes. In the UK many different professional groups and lay organizations are involved in health education and health promotion. 98

Health promotion in the UK

The Health Development Agency is the Department of Healths health promotion arm and succeeded the Health Education Agency in January 2000. Its website is http://www.hdaonline.org.uk/. The Agency is a special health authority. Its aim is to identify the evidence of what works to improve peoples health and reduce health inequalities. Then, in partnership with professionals, policy makers and practitioners, it will develop guidance and work across sectors to get evidence into practice. Members of the Board of the Authority are appointed by the Secretary of State for Health and include leading gures from health, associated professions, the media, education and related elds. Primary care trusts are also charged with improving the health of the population for which they are responsible. Most of their budgets are committed to the provision of personal health services, but some of their resources are allocated to health promotion. Often this is through specialist health promotion staff. These staff use a combination of health education and community support to target particular issues. They tend to concentrate on high-prole issues such as cervical cancer, HIV or heart disease. Voluntary bodies, such as the Royal Society for the Prevention of Accidents, the British Heart Foundation, Cancer UK or environmental groups such as Greenpeace and the Friends of the Earth are all active in health promotion. Their contribution to the provision of knowledge to individuals, inuence on public policy and help in reorientating the health services is increasingly recognized.

Health promotion programmes

There are many different health promotion programmes. Some leading examples of current activities are outlined below.

Health promotion and education Chapter 13 nicotine replacement therapy. This is another example of how the health service can begin to move from providing a curative approach to one where prevention and education is the goal. It is important to remember that most people start smoking when they are teenagers and thus strategies targeted at children have also been encouraged, for example getting local authorities to enforce the law on sales of cigarettes to the under 16s. Strategies to reduce smoking The UK has a long history of providing information about the dangers of smoking through government-funded campaigns, advice from general practitioners and health campaigns in schools. Punitive tax on tobacco is one public health policy, which has been shown to be effective in reducing smoking. A 10% rise in price has been associated with a 1% reduction in smoking. Banning the sale of cigarettes to children under the age of 16 years and the prohibition of smoking in certain public places are other examples of relevant legislative policies. The banning of advertising in countries such as Canada and New Zealand has been shown to reduce tobacco consumption, and the UK and Europe are now following suit. Many companies and hospitals have attempted to create healthier environments by the introduction of no-smoking policies. Some have also funded smoking cessation support for their staff. Cinemas, airlines and some restaurants now ban smoking. In March 2004 the Republic of Ireland passed legislation to ban smoking in public places such as pubs and resturants. Little is done to support voluntary organizations nancially in their campaigns against tobacco. A Canadian campaign involving health authorities, Action on Smoking and Health (ASH) and the Canadian Cancer Society demonstrated the effectiveness of combined action in achieving a ban on tobacco advertising in that country. One of the goals that general practitioners have been set as part of the National Service Framework on Cardiovascular Disease involves identifying the number of tobacco smokers within their practice. They can then refer them to smoking cessation clinics or prescribe supportive treatment such as
Increase the price of cigarettes Ban advertising Ban smoking in the work place and public places Identify and counsel current smokers Provide smoking cessation clinics Enforce the law on sales to children

Health promotion
Target areas include: Smoking Alcohol Nutrition Exercise Sexuality

Smoking

Alcohol
Alcohol abuse is of increasing concern. It is estimated that in the UK up to 40 000 deaths per year are alcohol related, including a signicant proportion of the 3500 road deaths. Cirrhosis of the liver is now four times more common in middle-aged men than it was in the 1970s. Public policies relating to alcohol include the imposition of excise duties and the passing of licensing laws. The UK has among the highest rates of tax on alcohol in the EU. The licensing laws were introduced initially to control the gin palaces of the 18th and 19th centuries. Paradoxically, these laws are now being relaxed. Another policy intervention aimed at reducing alcohol-related deaths was the passing of the drinkdriving laws. This has resulted in a considerable reduction in the number of deaths on the roads. Doctors have not always been good advocates or role models for the prevention of alcohol abuse. The tradition of medical student drinking can lead to the development of unhelpful professional and personal attitudes to drink. Strategies aimed at creating supportive environments to contain the abuse of alcohol should include offering people healthy choices, for example putting water on the table at mealtimes both in the home and when eating in restaurants. Offering food in pubs and other 99

Chapter 13 Health promotion and education places where alcohol is served also encourages more responsible drinking. Education includes giving people information about safer drinking levels and publicizing the existence of help agencies. Often, conicting information about the health benets of moderate drinking is preferentially heard, perhaps encouraging light drinkers to drink more whilst doing nothing to encourage the heavy drinker to reduce intake. Advice on dealing with alcohol abuse can be provided to individuals. To do this those people with a problem need to be identied. Simple screening questionnaires on all at-risk patients can be used both in hospital practice and in primary care. oils rather than animal fats. Whilst a population approach to nutrition is attractive, the use of a targeted approach in certain situations is also valuable. For example, preconception advice for women concerning their intake of folate will reduce the risk of them having a baby with a neural tube defect. Perhaps more could be done to improve nutrition through the adoption of nutritional policies. For instance, one initiative by the Department of Health has been the Five a Day programme which has been taken up by a number of primary care trusts and aims to get at-risk populations to eat ve portions of fruit and vegetables a day. The Government has also launched the Food in Schools programme which aims to improve school childrens knowledge about healthy nutrition. This programme was launched through the British Nutrition Foundation (http://www.nutrition.org.uk/). The other important body is the Scientic Advisory Committee on Nutrition (SACN). This is a UKwide advisory committee set up to provide advice on scientic aspects of nutrition and health. This includes advice on the nutrient content of individual foods and advice on diet as a whole including the denition of a balanced diet, and the nutritional status of people. They are also consulted on nutritional issues that affect wider public health policy issues including conditions where nutritional status is one of a number of risk factors (e.g. cardiovascular disease, cancer, osteoporosis and/or obesity). The website is http://www.sacn.gov.uk/.

Strategies to reduce harm from alcohol abuse

Increase the price of alcohol Drinkdriving laws Make water and soft drinks easily available Only offer alcohol with food Identify and counsel problem drinkers

Nutrition
The subject of nutrition is full of mixed messages, due to the paucity of consistent scientic evidence on the health effects of dietary change. In most parts of the world, malnutrition is the greatest threat to health. In the developed world, obesity is now a major problem. Public policy in the eld of nutrition has been scant and poorly coordinated. The Health of the Nation document published by the UK DoH in 1990 promoted a reduction in the percentage of food energy derived from fat and also aimed to reduce the prevalence of obesity. Despite this there has been a year-on-year increase in the prevalence of obesity. There are differential tax (VAT) rates on some foods, but legislation concerning food is generally aimed at minimizing known hazards rather than supporting nutritional objectives. Education about diet is widespread and often most effectively undertaken by food manufacturers, for example encouraging the consumption of cereals, and the choice of margarine or vegetable 100

Strategies to improve nutrition

Education through the media No tax on healthy foods Targeted messages, e.g. folic acid for pregnant women Scientic advice available to policy makers Introduce nutrition on the school curriculum

Exercise
The health benets of exercise are widely recognized and yet its promotion is often uncoordinated. This is one area where public policy could

Health promotion and education Chapter 13 have great inuence. Some new towns in the 1970s were designed with cycle paths and well-lit walkways to encourage healthy options for getting to and from work. The majority of local authorities have invested in sports facilities and made them available at subsidized rates, but many schools sold their sports grounds in the 1990s thus discouraging children from taking part in regular sports. Recently this has been counteracted by a new PE and Sports Programme funded through local authorities with the aim of increasing the provision and use of sports facilities. The Healthy Schools Programme has also emphasized the importance of physical activity to children. Knowledge about the benets of exercise has increased dramatically over the last two decades. This information is now being passed on by doctors to their patients. Patients may be referred to rehabilitation programmes, which increasingly emphasize the value of physical tness. Much of this activity is in the form of tertiary prevention, as after a heart attack. However recent randomized controlled trials have shown the benet of regular exercise as a primary prevention strategy to reduce the risk of developing diabetes. statements by the GMC and BMA about the prescribing of the pill to girls below the age of consent. The Government has a policy of providing free contraceptive services through general practitioners and family planning services, but ease of access to services has to be complemented by appropriate knowledge and behaviour. This is best encouraged through health education and by providing supportive environments. The change in attitude to the advertisement of condoms on television and their widespread availability through supermarkets and other retail outlets was brought about by a need to promote a change in behaviour to try to reduce the spread of HIV. This has had an effect on other STDs as well as making people more aware of the risks of unwanted pregnancy. This example shows how one health issue cannot always be separated from others. Some changes in health services seem to happen by accident. Making the oral contraceptive available only on a doctors prescription placed a clear responsibility on doctors, involving them in their patients sexual behaviour. General practitioners in particular accepted this responsibility so that now family planning advice is a major part of their work. The medicalization of contraception led doctors to become involved in a number of other initiatives such as cervical screening and well women clinics. The pill has thus been a very successful inuence in reorientating doctors towards providing preventive rather than curative health care.

Strategies to increase exercise

Healthy public policy, e.g. cycle tracks Increasing the provision of sports facilities Sports in schools programmes Exercise for high-risk patients, e.g. to prevent diabetes Part of rehabilitation programmes, e.g. after a heart attack

Ethics of health promotion

The ethics of health promotion can be approached using the four principles often used when considering individual care.

Sexually transmitted disease and unwanted pregnancy

Improving health through changes in sexual behaviour will help reduce the number of unwanted pregnancies and sexually transmitted diseases (STDs). The laws designed to prevent underage sexual intercourse do little to reduce the incidence of teenage pregnancies. This growing problem and the obvious need for contraception led to policy

Ethical principles
Rights and responsibilities Benecence Non-malecence Justice

101

Chapter 13 Health promotion and education A key conict arises between the goals of health promotion and the rights of individuals to personal autonomy. People working in health promotion sometimes seek restrictions on personal behaviour in the interests of the public good. This can lead to conict with a signicant sector of the public who wish to retain their autonomy of decision-making. Most agree that where the autonomy of others is threatened such as by drunk drivers on the road, it is reasonable for society to intervene. However, legislating against personal risk-taking is more controversial. There are no laws preventing mountaineering or bungee jumping, although there is legislation on the use of seat belts, which are only of benet to the individual concerned. Similarly, the use of certain drugs is illegal although they usually only directly affect the individual user. Thus, the law and public attitudes on these issues are not always consistent. In relation to benecence and non-malecence, in many situations the amount of good or the amount of harm that may arise from many health promotion initiatives is not known. This is not a reason for inaction, but the community is entitled to answers to allow it to make informed decisions. Often the initiative to mount a preventive health programme is undertaken without proper consultation with the community. This is contrary to the philosophy of health promotion, but is often due to ignorance on how to undertake community consultation. As far as justice is concerned, it could be argued that funds should only be spent when there is a good prospect of benet to the health of the public. This has been recognized by the Health Development Agency who have developed the HDA Evidence Base so that health promotion programmes of proven effectiveness can be pursued. With regard to the targeting of programmes the ethics of a population-based approach must also be considered in the context of the needs to reduce the inequities in health between the poor and the rich. These considerations suggest that all health promotion campaigns should at least be submitted to an ethical review before being implemented, and that a facility should be in place to re-examine the issues as the programme progresses.

102

Chapter 14 Control of infectious diseases

Introduction
An infectious or communicable disease is an illness caused by the transmission of a specic microbial agent (or its toxic products) to a susceptible host. The agents can be bacteria, viruses or parasites. The majority of microbes are harmless to humans. Some, although not universally pathogenic, are potentially dangerous and may cause disease in unusual circumstances. Caution is needed not to attribute a disease to an organism which happens to be present as a commensal or contaminant. There are many factors that determine whether or not biological agents result in the spread of disease in a population. They can be broadly divided into the presence of reservoirs of infection, the method of transmission, the susceptibility of the population or its individual members to the organism concerned, and the characteristics of the organism itself.

Human
The human population is the reservoir of infection in diseases such as measles and chickenpox. Were these organisms to be eliminated from humans, the diseases they cause would be eradicated in the same way that smallpox has been eradicated. However, due to their high infectivity and ease of transmission, these diseases are difcult to eliminate despite the use of mass vaccination programmes. In addition, some infections may be carried by non-symptomatic individuals who may transmit them to others. Asymptomatic carriers are often difcult to identify. Human carriers are of three types: healthy, convalescent or chronic. Healthy carriers are people who are colonized by a potentially pathogenic organism without any detectable illness, for example staphylococcal carriage in the anterior nares or in the axilla, or coliforms in the gut. Convalescent carriers are people who have recovered from the illness but who continue temporarily to excrete the organism, for example salmonellae in faeces. Chronic carriers are people who, while remaining clinically well, may carry and excrete organisms continuously or intermittently over a prolonged period, for example typhoid carriers in whom Salmonella typhi may remain in the gallbladder for life. Such carriers are a continuing threat to 103

Reservoirs of infection
A reservoir of infection is the site or sites in which a disease agent normally lives and reproduces. Reservoirs of infection may be classied as human, other biological or environmental.

Chapter 14 Control of infectious diseases the community long after they recover from the disease. Human immunodeciency virus (HIV) is of particular interest because the reservoir of infection is human. All carriers are infectious. Infectivity is at its highest around the time of seroconversion often when HIV infection has yet to be diagnosed and again later when HIV disease (the symptomatic phase) occurs.

Transmission survival
Organisms vary in their capacity to survive in the free state and to withstand adverse environmental conditions, for example heat, cold, dryness. Sporeforming organisms, such as tetanus bacilli which can survive for years in a dormant state, have a major advantage over an organism like the Gonococcus which survives for only a very short time outside the human host.

Other biological or environmental

These include: animals, for example Escherichia coli, rabies, malaria, psittacosis and hydatids; foodstuffs, for example Salmonella, Campylobacter and Listeria; water, for example giardiasis, schistosomiasis and cholera; soil and the environment, for example anthrax, Legionella, tetanus.

Life cycle
The life cycle of certain organisms has important consequences in the spread of disease. Organisms such as the malaria parasite which have a complex life cycle requiring a vector are more vulnerable than those with simpler requirements for transmission. In many infections by such organisms, humans are an accidental host.

Host susceptibility Transmission

Infectious diseases can be transmitted by various means and their mode of transmission inuences the spread of disease through a community. Interrupting the transmission of infectious agents is a key strategy for the control of these diseases. Methods of transmission include the following. Host factors that inuence the natural history of infectious diseases include the following.

Host factors
Age Gender Nutrition Genetics Immunity: natural, acquired and population

Transmission
Direct contact touching, kissing or sexual intercourse, e.g. Staphylococcus, Gonococcus and HIV Vertical transmission (mother to fetus), e.g. hepatitis B, Listeria, HIV, rubella and cytomegalovirus Inhalation of droplets containing the infectious agent, e.g. tuberculosis, measles, inuenza Ingestion of food or water that is contaminated, e.g. Salmonella, Giardia, Norwalk virus, hepatitis A Injection either by human interference or by insects, e.g. hepatitis B and C, tetanus, malaria

Age
The very young and the elderly are more susceptible to infectious diseases than are older children and younger adults. However, some common diseases of childhood such as measles, mumps and chickenpox can be more serious when they occur in adolescents and young adults.

Transmission is also affected by the conditions which organisms require for their survival and their life cycle. 104

Gender
There is some evidence that susceptibility to some infections differs with gender. In general, males ex-

Control of infectious diseases Chapter 14 perience higher age-specic mortality rates than females for most diseases.

Population (herd) immunity

The resistance of groups of people to the spread of infection is termed population (or herd) immunity. It depends on the proportion of individuals in the population who are immune. If this is sufciently high, chains of transmission of the agent cannot be sustained because susceptible people in the group are shielded from exposure to infected people by the immune people around them. The degree of herd immunity that will inhibit spread varies with different infections but is usually less than 100%. It depends on: the frequency of new introductions of infection; the degree of mixing which affects opportunities for contact between infected and susceptible people; and the transmissibility of the infection and duration of infectiousness of excreters. Herd immunity affects the periodicity of epidemics. So long as each case leads to more than one new infection, the incidence of the disease increases and herd immunity rises. When herd immunity reaches a level at which each case causes less than one new infection, incidence declines. As individual immunity wanes or new, susceptible people are introduced to the group, herd immunity again declines and the group is again vulnerable. This was well illustrated by the periodic epidemics of measles, which occurred every 23 years before the introduction of measles vaccination (see Fig. 3.4). Introduction of vaccination programmes lengthens the period between epidemics. The higher the immunization rate, the longer the period. If the antigenic composition of an infectious agent changes or if an agent previously absent from the population is introduced, there is no benet from herd immunity against that organism and large-scale epidemics may result. For example, antigenic changes of the inuenza virus from time to time lead to worldwide pandemics.

Nutrition
The state of nutrition of the host is very important. For example, in developing countries, measles may have a mortality of 5% amongst those who are poorly nourished whilst in the UK the case fatality rate is 0.02%. It is likely that the improvement in nutrition during the 19th century was a major reason for the reduction in deaths from communicable diseases at that time.

Genetics
Some individuals appear to have an exceptional susceptibility to infections, which is probably inherited. This can be seen in the similar susceptibilities of monozygotic twins and different susceptibilities of dizygotic twins to certain infections. In national or ethnic groups, natural selection over many generations may eventually breed a relatively resistant stock. A good example of this phenomenon is the history of tuberculosis in Europe. During the 19th century, the population experienced a high incidence of this disease which, by causing high mortality amongst susceptible young adults, tended to favour the survival through reproductive life of those with higher innate resistance. By contrast, when an infectious disease is rst introduced into a community with no prior experience of it, the result can be disastrous. For example, the introduction of measles to the Greenland Inuits by the American forces during the Second World War caused devastating epidemics with high mortality. Some genetic traits can be an advantage; for example, carriers of sickle-cell disease have a positive advantage when infected with malaria.

Immunity
The occurrence of disease in humans depends upon the individuals susceptibility to the agents to which he or she is exposed. Defence mechanisms are natural and acquired immunity (see Chapter 15) and population (herd) immunity.

Characteristics of the organism

The characteristics of the causal organism are also pertinent to the spread of infectious diseases. These include the following. 105

Chapter 14 Control of infectious diseases Some organisms are antigenically more potent than others. Those that invade the bloodstream, for example chickenpox, are more likely to produce a good immune response than those organisms that only infect surface membranes, for example the Gonococcus.

Organism characteristics
Infectivity: capacity to multiply in host Pathogenicity: capacity to cause disease in host Virulence: pathogenicity in a specic host Immunogenicity: capacity to induce specic and lasting immunity in host Antigenic stability: can induce lifelong immunity

Antigenic stability Infectivity

The infectivity of an organism is its capacity to multiply in or on the tissues of the host. This varies between microbial species, between individuals and with the route of entry. It may also be affected by the presence of tissue trauma, which facilitates the entry of organisms and provides a suitable growth medium. Organisms which are antigenically stable or exist in only one antigenic form, for example measles virus, usually induce lifelong immunity. If the agent is antigenically unstable, for example inuenza virus, or exists in many antigenic forms, for example rhinovirus, humans cannot develop lasting immunity. Environmental conditions, such as those created by the indiscriminate use of antimicrobial drugs, may select out the more virulent and resistant strains of bacteria from among several coexisting variants.

Pathogenicity
The pathogenicity of an organism is its capacity to cause disease in an infected host (i.e. ratio of number of cases of disease to total number of people infected). In the days before smallpox was eradicated, nearly every infection with smallpox virus in susceptible people caused disease (high pathogenicity), whereas many children infected with poliovirus are asymptomatic (low pathogenicity).

The environment and infection

The environment is the physical, biological and social world external to the individual. Environmental conditions interact in complex ways in facilitating the occurrence and spread of infection in human populations. For example, climate regulates the natural ora and fauna and the parasites that can survive and be transmitted. If the ambient temperature is warm, the multiplication of salmonellae in contaminated food is accelerated; malaria is transmitted only where the climate favours survival of Anopheles mosquitoes. Similarly the quality of housing, particularly the facilities for washing and waste disposal, inuences the transmission of infectious diseases and the presence of vectors. When sanitation is poor, epidemics of diseases such as cholera, plague, typhus and typhoid can soon appear. Improved transportation (whether road, rail or air) between communities has facilitated social intercourse and the spread of infective agents. Infection which spreads from person to person does so more rapidly where there is overcrowding, whether in army barracks, slum tenements or village communal huts.

Virulence
Virulence is the pathogenicity of an organism in a specic host. Different strains of the same agent may vary in virulence; for example, wild strains of measles and poliovirus are virulent in humans in contrast to the attenuated strains used in vaccines. The virulence of particular organisms may vary over time; for example, the virulence of Streptococcus pyogenes appears to have diminished over the last 80 years.

Immunogenicity
Immunogenicity is the capacity of an organism to induce specic and lasting immunity in the host. 106

Control of infectious diseases Chapter 14 The pattern of an epidemic depends on the biological properties of the agent, on whether or not the environment is favourable to its survival and transmission, and on the immunity of the host population. The course of an epidemic is therefore a reection of time, place and person interaction. Its investigation is an exercise in descriptive epidemiology. Epidemics are usually due to microbial agents although they can arise from other causes, such as chemical poisoning or mass psychogenic illness.

Control of infectious diseases

Some infectious diseases can have serious effects on the health of a population if they are allowed to spread unchecked. They may cause epidemics or the disease may become endemic.* In most western countries, such diseases are notiable by law to the public health authorities (see Table 8.1, p. 59, for list of infectious diseases notiable in the UK). As many of these diseases are food- or water-borne, the local government authority may be partly or wholly responsible for instituting environmental control measures. In other infections, control may be aided by use of vaccines and effective treatment of cases. Because of the numbers of people travelling around the world the transmission of diseases between countries is becoming an increasing problem. Severe acute respiratory syndrome (SARS) and West Nile fever are recent examples. Diseases that have originated or been endemic in one part of the world are rapidly transmitted to a virgin population. New measures are required to prevent such diseases being carried from one country to another.

Denitions
Before describing the different types of epidemics and outbreaks and their investigation it is necessary to explain some of the terms used (Fig. 14.1). Primary or index case(s) This is the rst case (or group of cases) arising from the introduction of an agent into a community. Secondary cases People who acquire infection from the primary/index case(s) are called secondary cases. Incubation period This is the interval between infection of an individual and the onset of symptoms. This is different for each organism and may vary for the same organism according to such factors as the virulence of the particular strain, the infecting dose and the susceptibility of the host. Serial interval/generation time This is the interval between the onset of primary and secondary cases. This interval may be shorter or longer than the incubation period depending on the duration of infectivity of the primary case, which may start well before and continue for some time after the onset of symptoms. When infection in intermediate cases is subclinical, the serial interval may be more prolonged than usual. Derived infection This is an infection arising by direct transmission from an infected contact. 107

Epidemics and outbreaks

The essential characteristic of an epidemic is that it involves a temporary increase in the incidence of a disease, usually circumscribed both in its location and in respect of the groups affected. Rarely, a worldwide epidemic of an infectious disease may occur (pandemic). The term outbreak is used to refer to the localized temporary increase in the incidence of a particular disease where the cases are potentially linked to each other. As few as two cases of a disease, associated in time and place, in circumstances where the disease is not a usual occurrence and/or a particular threat are sufcient to constitute an outbreak requiring investigation, for example meningococcal infection.
*An endemic infection is one that is usually present in a given geographical area or population group at relatively high prevalence and incidence rates in comparison with other areas or populations.

Chapter 14 Control of infectious diseases

Infection of case 1 (primary or index case) Incubation Symptoms Infectivity Initial exposure

Transmission from case 1 (primary) to case 2 (secondary)

Transmission from case 2 to case 3

Case 1

Case 2

Case 1

Case 2

Case 1 (4d)

Case 2 (5d) 5d

Case 3 (6d) 7d

Serial interval (generation time) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 Days Figure 14.1 Model of infectious disease transmission. d, days.

Secondary attack rate This is the number of new cases of a disease arising within one incubation period after the primary case(s). It can be expressed as: number of derived infections/number of susceptible persons in the group at risk.

Types of epidemic
There are two main types of epidemic: common source and propagated.

gates at a medical conference (Fig. 14.2). The vehicle by which the Salmonella was transmitted in this instance was contaminated chicken pieces served at a buffet lunch. The resulting gastrointestinal infections caused 196 doctors to report symptoms, of whom 32 were admitted to hospital. Over 1600 doctor-days were lost to the NHS. Example In 1996 the largest UK outbreak of E. coli O157 food poisoning occurred in Lanarkshire in Scotland. Over 500 cases were identied and 20 deaths resulted. The outbreak was traced to contaminated meat from a single butcher. The report into the outbreak highlighted concerns about food hygiene and the potential cross-contamination between raw meat and cooked meat products.

Common source epidemics

These epidemics result from the exposure of a group of people to the same source of infection or noxious substance. If exposure is simultaneous for all subjects, an explosive outbreak will occur one incubation period later and the duration of the epidemic will depend upon variation between individuals in the incubation period for the disease. Continuous or intermittent exposure of the population to the causal agent produces a more extended and irregular epidemic curve. The control of such outbreaks depends on the early detection of the cause and its removal at source. Example In 1986, there was an outbreak of Salmonella typhimurium food poisoning amongst dele108

Propagated epidemics
These are due to the transmission of the infectious agent from one person to another, for example measles or whooping cough. In such cases, the epidemic curve usually shows a gradual rise and decline, often with further waves as each successive generation of cases infects a new generation. The speed at which a propagated outbreak spreads depends on the interaction of a number of factors. These include the opportunity for contact

Control of infectious diseases Chapter 14

Figure 14.2 Number of cases according to time of onset. (From Palmer SR, Watkeys JEM, Zamiri I et al. J Roy Coll Phys Lond 1990; 24(1): 269.)

50 45 40 35 30 25 20 15 10 5 0

Number of cases

Buffet

12 5 Sept

12 6 Sept

12 7 Sept

12 8 Sept

12 9 Sept

Figure 14.3 Measles epidemic in a primary school. (From Graham R, Bellamy S, Richardson HJ. Commun Dis Rep 1979; number 16.)

16 14 12 10 8 6 4 2 1 3 5 February 7

Number of cases

Unvaccinated Vaccinated

9 11 13 15 17 19 21 23 25 27 1 3 March Measles cases by date of onset

between infected and susceptible people which is itself inuenced both by the density of population and by the level of herd immunity. Obviously, person-to-person spread is more likely to occur where large numbers of susceptible people are living in close proximity, particularly if there is a regular supply of new susceptible individuals joining the community, for example nurseries, schools, military camps, cruise ships, etc. Different organisms and different strains of the same organism may vary in their virulence, the speed at which they spread, the carriage rate in a particular community and the duration in individuals. Remote communities tend to be relatively protected by their isolation from some infections. However, once infection is introduced it is liable to spread with exceptional rapidity because herd immunity is usually low. For example, respiratory infections introduced into isolated island communities can cause very high morbidity rates. An

epidemic may be initiated from a common source and then continue by secondary spread from person to person. Example An outbreak of measles occurred in a primary school (Fig. 14.3). After two index cases in early February, there were two epidemic waves at approximately 1014-day intervals, i.e. the median incubation period for measles. The outbreak was modied by the fact that many of the children in the school had been vaccinated, including some who contracted the disease. The attack rate in unvaccinated children was high (86%) and showed the typical wave pattern of a propagated epidemic.

The investigation of outbreaks

Most epidemics are public health emergencies and require rapid and coordinated action to identify the cause and to institute effective control meas109

Chapter 14 Control of infectious diseases ures. It is wise to follow a systematic procedure in the investigation of outbreaks. subclinical infections are carried out. Phage, serological and other methods of typing of organisms may help to establish the epidemiological association between cases and possible causes (or sources) and to trace the paths of spread of the agent. Note The application of other epidemiological techniques such as the use of casecontrol studies may also be of value in the investigation of outbreaks as a means of conrming the validity of a causal hypothesis. In large outbreaks, investigations can sometimes be conned to random samples of patients and people thought to be at risk.

Outline of procedures
The steps described here are not necessarily undertaken in the sequence given. Enquiries usually proceed simultaneously with the analysis of ndings and often with interim control measures based on early indications of the likely origin of the outbreak. Not all the steps will be relevant in every outbreak and the questions asked must be adapted to the circumstances. The ve main stages in an investigation are shown below.

Investigation of reservoirs and vehicles of infection

Stages in investigation
Descriptive enquiries into the facts of the outbreak Investigation of reservoirs and vehicles of infection Analysis of the data collected Formulation of a causal hypothesis Testing its validity in the control of the outbreak

Human An epidemic may originate from an individual who has had a minor clinical episode or from a carrier who was ill many years previously. Therefore, a careful history should be taken from all contacts of the patients. Animal Enquire about the contacts patients may have had with sick animals or animal products known to harbour the infection concerned. Environment Investigate sources of foods consumed by affected individuals and the circumstances of their production, storage, preservation and preparation. Particular attention should be given to looking for situations in which cross-contamination or incubation of organisms could have occurred. Arrange for laboratory examination of food remnants, milk, and water supplies, and other relevant specimens from environmental sources, for example kitchen utensils, drains, etc., and the typing of any organisms that are isolated. Analysis of the data collected Plot the epidemic curve. This may give some clue to the mode of spread and probable time of initial exposure. For example, an outbreak of Salmonella napoli caused by contaminated chocolate bars im-

Descriptive enquiries Verify the diagnosis by clinical and laboratory investigation of the cases. Verify the existence of an epidemic by comparison with previous incidence of the disease in the same population. Compile a list of all cases and search for unreported cases by alerting hospitals and general practitioners in the district and neighbouring districts. Investigate patients and others who might be involved in the outbreak. Record the personal characteristics of the patients (age, sex, address, etc.) and enquire into shared experiences or activities that could carry risk of exposure to the suspected agent, for example occupation, school attended, recreational activities, consumption of foods, drugs, etc. Identify the total population at risk, i.e. all those who may have been exposed to the same hazards as the patients, whether ill or not. Ensure that all the clinical and laboratory investigations required to conrm the identity of the infection in patients and to determine the extent of
110

Control of infectious diseases Chapter 14 ported from Italy is shown in Fig. 14.4. Note the relationship between the time distribution of cases and the importation of bars of chocolate. Plot the cases on a map. This will detect clustering. The distribution of cases must be examined with reference to that of the population at risk. Analyse the incidence rates in different groups. This can be done, for example, for age or occupation. A high rate in a particular group suggests that the cause lies in a common experience of its members. Attack rates must be calculated both in those exposed and in those not exposed to the suspected agent. It should be noted that variations in the biological response to infection may result in clinical attack rates of less than 100% in the exposed population. Look for a quantitative relationship. This may exist between the degree of exposure (or dose) and attack rate, for example amount of suspect food consumed or closeness to a source of pollution. For example, in the outbreak of Salmonella typhimurium referred to under Common source epidemics (p. 108), food histories were obtained from 266 delegates at the suspect meal. Of these guests, 196 reported illness. The food-specic attack rates showed clearly that chicken was the probable vehicle of infection (Table 14.1).

Factors for hypothesis

The properties of the agent, its reservoirs and favoured vehicles and also of the nature of the illness it causes The probable source and route of transmission. For this purpose the typing of the organisms may be particularly helpful Time and duration of exposure of the patients to the agent in relation to the onset of their illness Attack rates of the different subgroups of the population at risk

Testing validity in the control of the outbreak Seek support for the causal hypothesis by further investigation of cases, if necessary, to conrm the proposed explanation of their illness. Carefully designed casecontrol studies may be very helpful in this. Implement appropriate control measures on the assumption that the hypothesis is correct and monitor their success in reducing the incidence of further cases.

Control of food-borne infection

The most frequently reported notiable infectious diseases are food poisoning and gastrointestinal infections. They illustrate well some of the biological and environmental factors that are conducive to the occurrence of outbreaks and the approach to their investigation and control outlined above.

Formulation of a causal hypothesis The hypothesis should take account of the following.
45 40 35 Number of cases 30 25 20 15 10 Figure 14.4 Number of cases of infection with Salmonella napoli from chocolate during AprilAugust 1982. (From Roberts JA, Sockett PN, Gill ON. Br Med J 1989; 289: 1227.) 5 0 April 4

Recall of chocolate and health warning 202 Primary household cases 43 Secondary cases Later importations of chocolate

March importation of chocolate

May 2

June 6 Date

July 4

August 1

September 5

111

Chapter 14 Control of infectious diseases

Table 14.1 Food poisoning attack rates for delegates eating and not eating specic foods. (From Palmer SR, Watkeys JEM, Zamiri I et al. J Roy Coll Phys Lond 1990; 24(1): 269.) Eaten Food Tuna Ham Beef Salmon Egg mayonnaise Pt Beef sandwiches Ham sandwiches Chicken Quiche (cheese) Quiche (ham) Ham and turkey pie *c2 = 70.7; P < 0.01. Ill 70 48 29 38 67 50 10 15 182 80 18 103 Total 98 63 46 46 89 66 13 20 213 108 21 137 % 72 77 64 84 76 76 79 75 86 50 86 76 Not eaten Ill 127 149 168 159 130 147 187 182 15 117 179 94 Total 169 204 221 221 178 201 254 247 54 159 246 130 % 76 73 76 72 73 74 75 74 29 74 73 73 RR 0.9 1.1 0.8 1.2 1.0 1.0 1.1 1.0 3.0* 0.7 1.2 1.0

They also exemplify the complementary roles of the health agencies and local authorities in the investigation and management of an outbreak.

Sources of contamination
Food may become polluted or infected at any stage during its manufacture and processing, distribution or preparation for consumption.

Causes of food poisoning

Food poisoning may be caused by either microorganisms or chemicals. In the case of microbiological food poisoning, the food may be either the vehicle whereby an agent is transmitted or the growth medium for the organisms. For example: salmonellosis may be caused by the organism being transmitted from poultry to humans in eggs; staphylococcal food poisoning may arise if during preparation the food becomes infected from a septic lesion in the food handler. If the food is then stored for long enough at a temperature which allows the organism to multiply, the toxins produced may result in severe symptoms of food poisoning in those who eat it. The harmful effects of chemicals may arise from either accidental contamination or the deliberate addition of chemicals to food as preservatives or in order to improve its taste or appearance. 112

Production
Salmonellosis usually owes its origin to the infection of livestock through their food or by crossinfection within herds or poultry ocks.

Manufacture and processing

In 1964 an outbreak of typhoid in Aberdeen was caused by corned beef which had probably become contaminated by use of polluted water to cool cans which had defective seals. The Lanarkshire outbreak of E. coli O157 noted above was due to contamination of cooked meat products prepared in a butchers shop.

Storage and distribution

Outbreaks of food poisoning due to a variety of agents have occurred because butchers, dairies and ice cream vendors have paid insufcient

Control of infectious diseases Chapter 14 attention to hygiene when storing and selling their products. premises and equipment, and on facilities for the storage and protection of food from contamination.

Preparation for consumption

In domestic households and in catering establishments, poor technique, particularly in relation to avoiding contact between raw and cooked meats, inadequate thawing of frozen foods, insufcient cooking and subsequent careful control of temperature during storage and serving, together with inadequate attention to cleanliness of premises and equipment, may lead to food poisoning, such as that due to Clostridium perfringens, staphylococcal toxins or Salmonella spp.

Education of food handlers

However strict the law, the avoidance of food poisoning depends heavily on those who prepare it. They should understand the importance of such matters as personal and kitchen hygiene in the avoidance of contamination or crosscontamination of foods. They should also appreciate the need, for example, to store food in protected containers and to adequately defrost frozen meat and poultry before cooking. The dangers of incubating organisms, especially in preprepared meat dishes, and the importance of refrigeration of foods liable to contamination in order to reduce bacterial growth and of the separation of raw meat from foods to be consumed without further cooking must also be constantly stressed.

Prevention of food-borne disease

The prevention of food-borne disease depends on correct action by many individuals in the complex chain of production, manufacture and distribution. The main ways in which the safety of food is maintained and good hygienic practice is encouraged are as follows.

Roles of CCDC and EHO

Cases of suspected food poisoning should be notied to the Consultant in Communicable Disease Control (CCDC) who are now employed by the Health Protection Agency (HPA). Their website is http://www.hpa.org.uk. The CCDC with the assistance of the EHOs employed by the Local Authority are responsible for the investigation of outbreaks of food poisoning. Outbreaks and single cases of serious infections, such as typhoid, call for immediate investigation and control measures. The results may call for amendment of food production, storage or preparation practices in the establishments concerned to avoid the danger of further episodes. In some cases it may be necessary to invoke legal powers to require replacement of faulty equipment, cleaning and refurbishment, or even closure of offending premises.

Quality of products
There are strict regulations relating to the quality and composition of some foods. This applies particularly to milk and milk products, meat and meat products, shellsh and the use of food additives by manufacturers.

Environmental conditions
Environmental health ofcers (EHOs) of local authorities have extensive powers to inspect all food premises and to sample foods. If necessary they can prevent their sale. The Food and Drugs Act (1955) and other relevant legislation laid down standards on the construction and cleanliness of food

113

Chapter 15 Immunization

Introduction
Historically, it was common knowledge that people who recovered from some infectious diseases, such as smallpox, rarely contracted that disease again. In 1796 Edward Jenner showed that a person who had been deliberately infected with cowpox was subsequently protected against smallpox. This led to the introduction of vaccination, one of the rst and most effective of all public health measures. The success of vaccination in eradicating smallpox from the UK and eventually from the world is well known. Discoveries at the end of the 19th century concerning the pathogenicity of bacteria led to the search for further vaccines. The isolation of anthrax by Koch in 1876 was quickly followed by Pasteurs (Fig. 15.1) attempts to develop attenuated strains that could be used to immunize animals and so protect them against the disease. Pasteur also developed an attenuated rabies virus that proved to be efcacious as a vaccine in humans. This was followed by other experiments, which showed that dead microbes, or their suitably modied toxic products (toxoids), could also provoke an effective immune response. In 1888, a diphtheria toxoid vaccine was developed. A successful vaccine against tuberculosis was not developed until 1921, an attenuated strain known as the bacille of Calmette and Gurin (BCG). During the Second World War, tetanus toxoid vaccine came into widespread use whilst an attenuated 114

virus vaccine against yellow fever provided protection for troops serving in the tropics. Today, we have available a great array of vaccines and new or improved vaccines are constantly being developed. The introduction of comprehensive immunization programmes utilizing vaccines against important diseases has done much to reduce mortality and morbidity worldwide, particularly amongst infants and children.

Passive immunization
Whilst most vaccines aim to induce lasting active immunity against specic infections, passive immunization can also be used to give short-term protection against a number of diseases. Passive immunization is the donation to the host of specic antibodies against a particular agent by the injection of blood products derived from immune animals or humans. It is used to give a degree of immediate, though temporary, protection to nonimmune individuals who have recently been exposed to a potentially dangerous infection. In such circumstances, active immunization may be of little benet because of the delay between administration of vaccine and the production of antibodies in protective amounts. Products used for passive immunization are immunoglobulins, which are now usually derived from the blood of human donors. The historical practice of using animal (usually horse) sera for

Immunization Chapter 15 there are a limited number of individuals who can donate their serum for the preparation of these products. Passive immunity to common infections occurs naturally through the transplacental transfer of antibodies from mother to baby. Similarly, antibodies are present in breast milk and give babies some protection against relevant infections while they are being breast-fed.

Active immunization
Active immunity to a disease is acquired naturally after recovery from infection with the causal organism. Articial active immunity can be induced by the administration of an appropriate vaccine which stimulates the production in the host of specic protective antibodies similar to those induced by natural infection. This provides complete or partial protection, usually lasting at least for a few years and in some cases for life. Active immunization is usually given as a planned procedure. It is designed both to protect individuals against infections to which they may be exposed at some time in the future and to control the spread of infection in the community (population (herd) immunity, see p. 105). While some types of vaccine produce a prompt and effective response after a single dose, the production of antibodies after the rst dose of other types of vaccine can be slow and inadequate. Multiple doses at intervals of days or weeks may be required to achieve protective levels of antibody. Further reinforcing doses at intervals may be necessary to maintain immunity in later life. Such doses (or later natural infection) stimulate an antibody response which is always more rapid and usually greater and more durable than the primary response.

Figure 15.1 Louis Pasteur (182295), chemist and originator of rabies vaccine.

this purpose has generally been abandoned because of the risk of anaphylaxis. The degree and duration of the protection afforded depends on the amount of antibody present, but signicant protection usually lasts no more than 36 months. There are two main types of immunoglobulin in use: human normal immunoglobulin and specic immunoglobulin. Human normal immunoglobulin is extracted from the pooled plasma of blood donors. This confers short-term protection against a range of infections that are either endemic or for which immunization is routine practice in the donor population, for example measles and hepatitis A. Specic immunoglobulin is prepared from the serum of individuals who have recently had a particular disease or have recently been actively immunized against the infection. Immunoglobulins of this type are prepared for varicella (chickenpox), tetanus, rabies, hepatitis B and a number of other infections. These tend to be in short supply and their use is carefully controlled. This is because

Types of vaccine
Vaccines are of four main types.

115

Chapter 15 Immunization uenzae type b (Hib) vaccine, which is prepared from puried capsular polysaccharide. Also acellular pertussis vaccine is now used in preference to the killed vaccine. Vaccines vary in their antigenic potency, i.e. their capacity to induce the formation of protective antibody. Much current work on vaccine development is focusing on producing vaccines that will produce a better immune response in a shorter time. One way of doing this has been particularly effective when producing vaccines for bacteria that have a protective polysaccharide capsule. Traditional vaccines have used simple capsular polysaccharides, but these vaccines have not been effective in infants, and have not provided longterm immunity. Attaching these polysaccharides to larger, more antigenic molecules to produce conjugate vaccines may overcome these problems. Antigenic potency can sometimes also be enhanced by the use of adjuvants such as aluminium phosphate or aluminium hydroxide which are included in the pentavalent diphtheria, tetanus, acellular pertussis, Hib, IPV vaccine.

Vaccine types
Inactivated or killed vaccines Live vaccines Toxoids Component vaccines

Inactivated vaccines
These are made from whole organisms, which are killed during manufacture. Examples include injected polio vaccine (IPV), typhoid, cholera and some pertussis vaccines.

Live vaccines
These are made from living organisms, which are either the organisms that cause the disease whose virulence has been reduced by attenuation (e.g. oral polio, measles, mumps and rubella vaccines) or organisms of a species antigenically related to the causal agent but which are naturally less virulent (e.g. smallpox (vaccinia) and tuberculosis (BCG) vaccines). In susceptible (non-immune) individuals these attenuated organisms multiply in the body to many times the quantity given in the original dose, but in an immune individual the virus is killed before it has a chance to replicate, so having little if any effect. This explains why it is believed live virus vaccines including measles, mumps, rubella and polio can safely be repeated in people who have been vaccinated previously.

Site of vaccinations
The route of administration varies between vaccines. Most are injected, whilst some are given orally. The site of the injection is important for two reasons. Firstly, the antibody response varies depending on whether the injection is given intramuscularly, subcutaneously or intradermally. Secondly, the frequency of adverse effects varies from site to site. Some vaccines, if given too deeply, can cause severe reactions. For example, BCG vaccine must always be given intradermally and should only be given by trained vaccinators. Live polio vaccine is given orally which has the advantage of stimulating local immunity in the intestine and inhibits later colonization (and transmission) of wild poliovirus. Most other vaccines are normally given by intramuscular or deep subcutaneous injection. In infants, the recommended sites are the anterolateral aspect of the thigh or upper arm. If the buttock is used, the injection should be into the upper outer quadrant to avoid the risk of sciatic nerve damage.

Toxoids
These are produced from bacterial toxins articially rendered harmless (e.g. diphtheria and tetanus toxoids).

Component vaccines
These contain one or more of the component antigens of the target organism that are necessary to provoke an appropriate protective antibody response. Examples of component vaccines, sometimes called subunit vaccines, include inuenza and hepatitis B virus vaccines and Haemophilus in116

Immunization Chapter 15 In order to reduce the number of separate injections, several agents are sometimes incorporated in the same vaccine. For example, the pentavalent vaccine for infants contains diphtheria, tetanus, acellular pertussis, Hib, IPV vaccine whilst MMR includes measles, mumps and rubella vaccines. When giving more than one live vaccine it is considered advisable to give them on the same day in different sites (unless an approved combined preparation is used) or to separate them by an interval of not less than 3 weeks to improve the immune response. ing ts and irreversible brain damage. More recently there has been concern about the measles vaccine causing inammatory bowel disease, and lately MMR has been linked to the increase in autism. Despite scientic evidence that these risks are small or non-existent, the impact these scares have on immunization rates can be dramatic and are a threat to the public health.

Anaphylaxis
Anaphylactic shock after vaccination is much feared and can be life-threatening, but it is very rare. In the 3 years from June 1992 there were 87 spontaneous reports of anaphylaxis and no deaths. Over the same period 55 million doses of vaccine were supplied in the UK. Thus the probability of a vaccinator encountering a case of anaphylaxis is very small. Nevertheless, adrenaline and appropriate airways should always be at hand and all doctors and nurses responsible for immunization must be familiar with the management of an anaphylactic reaction.

Safety and efcacy of vaccines

No new vaccine is released without extensive safety tests in animals and controlled eld trials designed to establish the level of efcacy and expected nature and frequency of adverse events after vaccination. Careful observance of specic contraindications to each vaccine reduces the risk. Nevertheless, some vaccines frequently give rise to minor reactions, for example local oedema at the injection site, transient fever or rash. Serious systemic reactions, especially neurological conditions, cause great concern but are very rare. To assess their signicance, routine surveillance must be maintained. Careful records should be kept of all the vaccinations given, to whom and where, with particulars of the vaccine used. Any serious reactions should be reported at once to the Committee on Safety of Medicines (on a Yellow Card). Likewise, the continued efcacy of a vaccine in controlling a disease should be monitored by the analysis of routine morbidity and mortality reports supported, where appropriate, by microbiological data and antibody surveys. In the UK, these studies are undertaken by the Communicable Disease Surveillance Centre (CDSC) of the Health Protection Agency. From time to time the safety of a vaccine comes under particular scrutiny. This is more likely to be an issue as the danger of the disease in question fades from consciousness whilst concerns about safety become relatively more important when considering risk and benet. Thus in 1976 there was concern about the pertussis component of the DTP triple vaccine with reports of children suffer-

General contraindications to vaccination

Immunization should be postponed if the recipient has a current acute or febrile illness. Immunization should not be carried out in an individual who has a history of a severe local or general reaction to a preceding dose. Live vaccines should not be given to pregnant women. Live vaccines should not be given to patients on immunosuppressive treatment or with immunosuppression due to disease. Live vaccines should not be given for at least 3 months after a dose of immunoglobulin or a blood transfusion.

False contraindications to vaccination

Prematurity. Infants who were born prematurely should be vaccinated at the recommended ages, i.e. 2 months, 3 months, etc. A previous episode of or contact with the disease concerned, for example measles or whooping 117

Chapter 15 Immunization cough, is not a contraindication because antibody testing has shown that the clinical diagnosis is frequently incorrect. There is no increased likelihood of complications following vaccination in those who already have natural immunity. Mild illness or chronic disease, for example asthma, diabetes. Mother or household member pregnant. A stable neurological condition. Family history of convulsions or adverse reactions. History of allergy except hypersensitivity to egg. ts of the vaccine their child is being given. Parents should be provided with written information and given opportunities to discuss their concerns.

Routine immunization
The current schedule for routine immunization recommended in the UK is shown in Table 15.1. The exact timing of doses is open to variation. While the ages recommended for each vaccine are considered to be optimum, it is important to ensure as far as possible that all children are vaccinated even if they present outside the recommended age range, unless there are specic contraindications (see Immunisation Against Infectious Disease, HMSO, 1996). More up-to-date information about the immunization schedule can be obtained from the website www.immunisation.org.uk.

Cold chain
Appropriate storage conditions are important, particularly for live vaccines, which need to be kept cold. Failure to maintain a cold chain during transport and storage may reduce the efcacy of a vaccine. The most common problem is the storage facilities in many doctors surgeries, where the constant use of refrigerators for other purposes may mean that the required low temperatures are not maintained.

Diphtheria, tetanus, pertussis, Hib and polio vaccines

In the UK it is recommended that primary immunization with diphtheria, tetanus, acellular pertussis, Haemophilus inuenzae type b (Hib) and inactivated polio vaccine should begin at the age of 2 months and be completed by 4 months. This is now done using a single pentavalent combination vaccine. This ensures protection against these

Consent
Informed consent should be obtained before each vaccination is given. This need not be in writing but parents should understand the risks and bene-

Table 15.1 Schedule of routine childhood immunization in the UK. Vaccine DTaP/Hib/IPV 2nd 3rd MMR 2nd DTaP/APV BCG Tetanus/IPV Dose 1st 3 months 4 months 1st 4 years* Booster 1st Booster Age 2 months

1224 months

5 years 1014 years (or may be given at birth) 1518 years (school leaving)

* A further routine dose of MMR at age 4 years has the advantage of boosting immunity in those who responded poorly to the rst dose and of protecting those who escaped a rst dose at 1224 months. Sometimes the second dose of MMR is given 3 months after the rst dose. BCG, bacille CalmetteGurin; DTP, diphtheria, tetanus, pertussis; Hib, Haemophilus inuenzae b; MMR, measles, mumps, rubella.

118

Immunization Chapter 15 diseases as early in infancy as possible. Fears about the safety of pertussis vaccine are now largely discounted and in any case probably only applied to the whole cell vaccine that was used previously. Reinforcing doses of diphtheria, tetanus, acellular pertussis and IPV should be given at or shortly before school entry. Further doses of tetanus, diphtheria and IPV are required at 1518 years. ent as a commensal organism of the nose and throat, it can cause pharyngeal inammation. Certain types of C. diphtheriae produce toxins, which cause the exudation of the classical pharyngeal membrane covering the fauces. The toxins produced can also cause cardiac failure and death. The bacterium is passed from person to person by direct contact or inhalation of infected droplets and is more common in young people. Thus, children living in overcrowded housing are particularly susceptible. Epidemics of diphtheria were particularly common in the 19th and early 20th century and caused the deaths of large numbers of infants and young children. Prior to the Second World War, there were around 50 000 notications each year and 3000 deaths despite the fact that a vaccine made from the toxin had been available since the 1920s. The death rate fell dramatically during the war years with the wider use of vaccine, and by 1954 the annual number of deaths was in single gures. Diphtheria is no longer endemic in the UK and the risk of infection derives only from imported cases or travellers to endemic regions.

Tetanus
Tetanus has been known to affect humans for centuries. The disease is caused by the circulation of neurotoxins that have been produced by the bacterium Clostridium tetani. The toxins cause severe muscle spasms which are extremely painful and may last for a matter of seconds, or continue for many minutes. As well as causing spasm of the jaw muscles (hence its common name lockjaw), increasingly persistent spasms cause respiratory failure and death. Clostridium tetani is found as a commensal in the large bowel of many animal species, including humans. The bacterium can form spores that are able to exist in a dormant state in soil for many decades and when introduced into the body by means of a contaminated penetrating wound may cause local infection with production and release of neurotoxins. A vaccine derived from the tetanus toxin was developed in the 1930s and was administered to millions of soldiers in the Second World War with great success. Today, tetanus vaccination is offered to all infants, with booster doses at 5 years and at school-leaving age. A reinforcing dose of tetanus vaccine may be required after certain types of high-risk injury or burns in individuals who were immunized more than 10 years previously. Where an individual with such an injury has no clear history of having completed a primary course of tetanus immunization, a dose of human antitetanus immunoglobulin should be given in a different site at the same time as the rst dose of a primary course of active immunization.

Pertussis (whooping cough)

Whooping cough was described in 1670 by Thomas Sydenham who called it infantum pertussis (violent cough of children). The Chinese described it as the hundred-days cough. It is caused by the highly infectious bacterium Bordetella pertussis and is spread by droplet infection. There is a catarrhal stage for 12 weeks before paroxysmal coughing develops. In young infants, the characteristic whoop may not be heard and coughing spasms may be followed by periods of apnoea. Complications of whooping cough include pneumonia, post-tussive vomiting, convulsions, and cerebral anoxia with a risk of brain damage. Most deaths occur in children under 6 months of age. In the UK in the past, whooping cough epidemics were seen every 35 years. Reduced vaccine uptake in the mid 1970s following concerns about the safety of the vaccine led to an increase in the incidence of pertussis, but this has been reversed following much improved vaccine uptake rates 119

Diphtheria
Diphtheria is a disease caused by the bacterium Corynebacterium diphtheriae. Although often pres-

Chapter 15 Immunization and increased population immunity in the last few years (Fig. 15.2). The whooping cough or pertussis vaccine is a component of the pentavalent DTaP, Hib IPV vaccine given at 2, 3 and 4 months. It is an acellular vaccine produced by inducing antigens to various relevant proteins. Concern that the killed Bordetella pertussis vaccine might cause brain damage was allayed following the National Childhood Encephalopathy Study (p. 43) which showed that the risk, if any, was extremely small in relation to the risk of disease. Children who have had a severe reaction to a previous dose should not have another dose and children with a developing neurological illness should also not be vaccinated. In these situations further advice should be sought. the rst conjugate vaccine to be licensed in the UK and was introduced into the immunization schedule in 1992, with three doses given at 2, 3 and 4 months of age. In addition, a catch-up programme was arranged for children up to the age of 4 years. Since then there has been a rapid reduction in morbidity and mortality due to this important pathogen (Fig. 15.3).

Poliomyelitis
Poliomyelitis was rst recognized as a distinct disease in the early 19th century and became known

600 500 Notifications

Hib vaccine introduced

Haemophilus inuenzae type b (Hib)

Haemophilus inuenzae is a common bacterium, which has a number of antigenic types. It is the H. inuenzae type b (Hib) which is the cause of nearly all invasive and life-threatening infections, particularly in children under the age of 5 years. It is a major cause of meningitis, with a case fatality rate of around 5%, and also causes life-threatening epiglottitis in young children. The Hib vaccine, rst produced in the 1970s, contains puried capsular polysaccharide conjugated to a protein. It was

Notificati

ons

400 300 200 100 0 1989 < 1 year 1990 1991 1992 Year (198994) 1993 1994
1 4 years

Figure 15.3 Notications of Haemophilus inuenzae type b vaccine (Hib), 198994. (Reproduced with permission of the OPCS (Crown copyright).)

200 000

Immunization introduced

81% Vaccine uptake

92%

150 000 Cases

Cases

2500 Deaths Figure 15.2 Whooping cough notications: cases and deaths in England and Wales, 194090. (Reproduced with permission of the OPCS (Crown copyright).) 2000

100 000 30% 50 000 Deaths 0 1940 1950 1960 1970 Year (194090) 1980 1990

1500 1000 500 0

120

Immunization Chapter 15 as infantile paralysis because it affected mainly infants and young children. The rst epidemic was described in Sweden in 1887. Major epidemics occurred in the UK during the late 1940s and early 1950s (Fig. 15.4). The rst vaccine developed against polio was inactivated virus (Salk) injected vaccine (IPV) which was introduced for routine immunization in the UK in 1956. It was replaced by the live attenuated virus (Sabin) oral vaccine (OPV) in 1962. Three types of poliovirus are included in both the oral and killed vaccines. Polio is frequently asymptomatic but can cause aseptic meningitis, severe paralysis and death. Paralysis may be as rare as one in 1000 infections in children and one in 75 in adults. Case fatality in people with paralysis varies from one in 50 in young children to one in 10 in older patients. The IPV (Salk) vaccine prevents the disease in vaccinated individuals but is less effective than OPV in creating population immunity because it reduces but does not prevent carriage of the virus in the bowel. The OPV (Sabin) vaccine contains live attenuated virus, which provides individual protection and also limits carriage and therefore transmission of wild virus. Very rarely the disease has been reported in vaccine recipients or in their non-immune contacts. Vaccine strains of poliovirus may be excreted for up to 6 weeks after vaccination. For this reason, oral vaccine has been replaced by IPV to immunize children. Adults who have not been immunized against polio in childhood should receive a primary course: no adult should be left unprotected against polio. Further reinforcing doses after that given routinely at 1518 years are not usually required except for travellers to countries where the disease is epidemic or endemic and for health care workers in contact with possible cases of the disease. In addition to the standard general contraindications to vaccination, polio vaccination should be postponed in patients with vomiting or diarrhoea.

Measles, mumps and rubella vaccine (MMR)

Measles
Measles is an acute viral illness, which is highly infectious in unvaccinated children. Before the vaccine was introduced in 1968, annual notications varied from 160 000 to 800 000 with peaks every 2 years (see Fig. 3.4). Since then, rates have declined with smaller and less frequent epidemics (Fig. 15.5). Complications occur in one in 15 reported cases and include convulsions and encephalitis, otitis media, pneumonia and bronchitis. Measles is thus, potentially, a major cause of acute and chronic ill health in children. Severe illness and death are more common in poorly nourished children and those with chronic conditions, but more than half the deaths occurred in previously healthy children. The vaccine is usually given shortly after the rst birthday. Earlier administration is not advised because the presence of maternal antibody may interfere with the active immune response. Unless a very high proportion of infants are immunized and develop a satisfactory response, there is a danger of accumulation of sufcient numbers of susceptible older children to sustain an epidemic. To reduce this risk a second dose is given with the preschool booster vaccinations.

10 000 Notifications 8 000 6 000 4 000 2 000 0

1931 36 41 46 51 56 61 66 71 76 81 86 91

IPV OPV

Year (193191)

Mumps
Figure 15.4 Polio notication in the UK showing the introduction of injected polio vaccine and oral polio vaccine, England and Wales, 193192. (Reproduced with permission of the OPCS (Crown copyright).)

Mumps is a common but not normally serious illness. However, complications including pancreatitis, oophoritis or orchitis (leading on occa121

Chapter 15 Immunization

600 000 500 000 Notifications 400 000 300 000 200 000 100 000 0 1965

Measles vaccine introduced Deaths Notifications MMR

140 120 100 80 60 40 20 0 Deaths Figure 15.5 Measles notications and deaths following the introduction of mass immunization for measles in 1968 and measles, mumps and rubella (MMR). (Reproduced with permission of the OPCS (Crown copyright).) Terminations

1970

1975

1980

1985

1990

Year (196594)

100 80 60 CRS CRS MMR

1200 1000 800 600

40 400 20 Terminations 0 1971 1976 1981 Year (197186) 1986 0 200

Figure 15.6 Numbers of terminations of pregnancies and births with congenital rubella syndrome (CRS) following the introduction of vaccine for rubella for girls in 1970 and measles, mumps and rubella (MMR) vaccine for boys and girls in 1988. (Reproduced with permission of the OPCS (Crown copyright).)

sion to sterility), meningitis and encephalitis can occur and justify the use of vaccine to prevent infection.

Rubella
Whilst rubella is a mild disease, maternal rubella infection in the rst 810 weeks of pregnancy results in fetal damage in up to 90% of infants and multiple defects are common. The risk of damage declines to about 1020% by 16 weeks gestation after which fetal damage is rare. Rubella vaccine was introduced in the UK in 1970 and was recommended for all girls aged between 10 and 14 years of age and for non-pregnant seronegative women of child-bearing age. The application of this policy over the years since 1970 has led to a fall in the number of conrmed rubella infections in preg122

nant women and with this the number of rubellaassociated terminations of pregnancy. As a consequence, the numbers of children born with congenital rubella syndrome also declined (Fig. 15.6). However, the selective vaccination of only girls and women allowed continued circulation of wild rubella virus in the community with the concomitant risk that a few women who had evaded immunization, or had failed to mount an adequate antibody response to the vaccine, could be exposed to infection in early pregnancy. Since 1988, when MMR vaccine was introduced, both boys and girls have been offered vaccination against measles, mumps and rubella in early childhood. This resulted in the virtual elimination of congenital rubella syndrome. The ultimate aim is to eliminate measles, mumps, rubella and congenital rubella syndrome. The routine vaccination of girls

Immunization Chapter 15 between the age of 10 and 14 years has now been abandoned but seronegative non-pregnant women of child-bearing age should continue to be given single antigen rubella vaccine. Despite the recent scare that MMR might cause autism it is believed that the combined vaccine is safe and more effective than using single vaccines. ease in humans. Mycobacterium tuberculosis was a major cause of morbidity and death in the 19th and early 20th centuries. There were over 20 000 deaths a year still occurring in the UK in the 1940s. It is an organism that usually causes infection of the lung or associated lymph nodes (pulmonary tuberculosis), although it can affect any part of the body (extrapulmonary tuberculosis). Respiratory infection can lead to localized disease, which is short lived and gives immunity to the individual, or it may cause progressive lung disease. Transmission of M. tuberculosis is normally by inhalation of air-borne droplets containing bacilli. The infection is more common when people are living in overcrowded conditions. The disease is also more common when the population is poorly nourished or has a high prevalence of chronic diseases. The death rate in the UK from M. tuberculosis has been decreasing steadily since the mid-19th century, the reduction being due principally to improved nutrition and living conditions. The advent of effective drug treatment and the widespread use of BCG vaccination accelerated the reduction (see p. 16). Notications of new cases of tuberculosis reached a low point in 1987. Since then there has been a small rise in the number of new cases (in 1992 there were 5798 notications) whilst the number of deaths each year is about 400. The rise in tuberculosis in the UK is mainly in the immigrant population and in the homeless. Developed in 1921, BCG vaccination was not introduced into general use in the UK until 1953. The routine use of BCG is controversial. Studies in different countries have produced conicting evidence of efcacy, the reasons for which are not clear. As a result, whilst it is accepted for routine use in some countries, others have not regarded its benets as proven and in some, where the incidence of tuberculosis has declined to the extent that it is no longer seen as cost effective, it has been discontinued. In the UK, BCG vaccine is given as a routine to school children at age 1014 years. It is also recommended for tuberculin-negative people in the following categories. Contacts of cases known to be suffering from active respiratory tuberculosis. 123

Meningococcus group C
Neisseria meningitides, the Meningococcus, is commonly carried in the nasopharynx. In some individuals, for reasons which are incompletely understood, it can become virulent, and can cause septicaemia, meningitis or a combination of the two. There are a number of antigenically different strains of Meningococcus, the most important strains being referred to as serogroups A, B, C, W135 and Y. The commonest of these in the UK has always been group B Meningococcus. Group C Meningococcus started to represent an increasing proportion of cases of meningitis, and seemed to be slightly more virulent than group B. Over a 5year period from July 1993 there were an estimated 3151 cases of group C meningococcal disease, mainly in young children and teenagers, causing 398 deaths and 1768 ITU admissions. Most deaths and ITU admissions occurred in teenagers aged 1519 years. Work is under way to develop a vaccine for serogroup B Meningococcus, but at the time of writing none is available. Polysaccharide vaccines have been available for the other strains mentioned for some time. These work for a relatively short time, are ineffective in younger children and do not prevent carriage (and therefore do not induce herd immunity), so they are not suitable for routine use. In 1999 a new group C conjugate vaccine was licensed in the UK, with none of the shortcomings of the polysaccharide vaccine, and this is now routinely given to babies with the primary course of DTaP, Hib, IPV.

Tuberculosis vaccine (BCG)

Mycobacterium tuberculosis is present throughout the world, including the UK. Other Mycobacterium species are also found and occasionally cause dis-

Chapter 15 Immunization Infants and children of immigrants in whose communities there is a high incidence of tuberculosis, who for this purpose may be regarded as contacts. (Newborn babies who are contacts need not be tested for tuberculin sensitivity but should be vaccinated without delay.) Health service staff. This category should include doctors, medical students, nurses and any other staff who may come into contact with patients or infected specimens from them. It is particularly important to test staff working in maternity and paediatric departments. The vaccine should not be given to tuberculin test-positive people because of the risk of severe reactions. lation (herd) immunity and to prevent the predicted epidemic. This strategy was effective in the short term but suggests the need to maintain a programme of preschool booster immunization.

Other vaccinations
Hepatitis B
Infection with the hepatitis B virus can cause disease ranging from a subclinical disturbance of liver function, to acute liver necrosis and death. The virus is transmitted by blood and semen. Some individuals may become chronic carriers, and these individuals are at increased risk of hepatocellular carcinoma. In some countries in south-east Asia the virus is endemic, there are many carriers and hepatocellular carcinoma is a common cause of death. Those infected by vertical transmission from mother to baby, or those infected at a very young age are much more likely to become carriers. In adults, acute liver failure is more common than in children but chronic carriage occurs in only 1% of cases. Hepatitis B vaccine is produced through recombinant DNA techniques. The vaccine is about 90% effective overall; it is slightly less effective in those over 40 years of age. The duration of vaccine-induced immunity is thought to be 35 years. It is recommended for doctors, dentists, nurses, midwives, laboratory workers, mortuary technicians, renal dialysis patients, the sexual partners of hepatitis B carriers and infants whose mothers are carriers. Parenteral drug abusers, prostitutes and other sexually promiscuous individuals of both sexes, morticians and embalmers, inmates of long-term custodial institutions, travellers to areas of the world where the disease is endemic and certain members of the police and other emergency services judged to be at high risk may also be considered for vaccination.

WHO immunization targets

The WHO Health for All by the Year 2000 targets announced by the European Ofce stated that: By the year 2000 there should be no indigenous poliomyelitis, neonatal tetanus, diphtheria, measles or congenital rubella syndrome in the European Region. The DoH in the UK supported this target and also included mumps and pertussis. To help achieve this, in 1985 the Government set a national target of 90% immunization rate for children under the age of 2 years. The Health of the Nation programme (1992) revised this to a target of 95% by 1995. Incentives were offered to general practitioners to achieve these targets which generally have been successful. Most UK regions were reporting immunization rates of 9095% by 1995. However, the targets are more difcult to achieve and sustain in inner cities, and other areas where there is a very mobile population. The Government also set a target of a 90% reduction in the number of notications of measles by 1995 compared with around 25 000 notied cases in 1989 (after the introduction of MMR in 1988). By 1994, the number had fallen to around 10 000 cases, but the relatively low historical immunization rates and modest but signicant vaccine failure rates left a substantial pool of susceptible individuals. This led to predictions of a large outbreak in 1995. In 1994, the DoH therefore instituted a catch-up programme aimed at school-aged children, to try to improve the popu124

Inuenza
Inuenza is an acute viral respiratory illness that usually occurs in epidemics during winter months. In healthy individuals, it is normally a mild illness,

Immunization Chapter 15 but can cause signicant excess mortality in the elderly and other vulnerable groups. Unpredictable changes in the virus surface antigens, which may partially or wholly invalidate immunity acquired from exposure to earlier variants, account for the irregularity of epidemics. If the antigenic shift is substantial, pandemics, sometimes with high fatality rates, may occur. There are two main types of inuenza virus, A and B, each of which can independently cause epidemics. Killed virus vaccines against both types have been shown to be protective. However, because of the antigenic instability of the inuenza virus, the value of the vaccine is variable and unpredictable. Vaccine is prepared from the latest antigenic variants of inuenza A and B virus, issued by the WHO. These are for use in the early autumn for people at special risk, such as the elderly (especially those living in residential institutions), and for those suffering from certain chronic diseases including pulmonary, cardiac and renal disease, diabetes and other endocrine disorders and conditions requiring immunosuppressive therapy. The vaccine is not recommended for the control of outbreaks. Live inuenza vaccines are still experimental and are not in general use in the UK. protection they require depends both on the country to be visited and also on the likelihood of their exposure. Thus, tourists staying in modern urban facilities are at much less risk from many diseases compared to an aid worker or backpacker who may be living or travelling for extensive periods in remote parts where serious infections are endemic and living conditions are poor. Health advice should include both general protective measures and advice on specic vaccinations. Diseases for which vaccinations are available include those passed via the oral/faecal route (hepatitis A, typhoid, cholera, polio), those spread by inhalation (tuberculosis, meningococcal diseases, inuenza), those passed by mosquitoes (yellow fever, Japanese encephalitis) and others such as rabies. Protection against diseases passed by the oral/faecal route depends principally on good personal hygiene and the avoidance of potentially contaminated food and water.

Typhoid
Vaccination is of value to those who are going to a country where they may have prolonged exposure to potentially hazardous food and water. Both a killed whole-cell vaccine and a live attenuated oral vaccine are now available and will give 7080% protection. Under conditions of continued or repeated exposure to infection a reinforcing dose should be given every 3 years.

Pneumococcus
Streptococcus Pneumoniae (the Pneumococcus) can cause pneumonia, septicaemia, meningitis or other infections. It is a major cause of illness, especially in the very young, the elderly, and those with an absent or non-functioning spleen or other causes of impaired immunity. There are well over 80 antigenically different strains. Two-thirds of the serious infections in adults and 85% of infections in children are caused by just 810 capsular types. People at higher risk should be vaccinated. Current vaccines include a polysaccharide vaccine, which covers 23 of the capsular types, and a conjugate vaccine, which covers nine capsular types.

Cholera
Cholera vaccine gives only limited protection (at most 50%) and is not considered to be of value in epidemic situations. Its use is therefore no longer recommended and it is no longer a legal requirement for entry to any country. The principal need in cases of cholera is for adequate rehydration. If properly managed, cholera is rarely lifethreatening in those who are well nourished.

Vaccination for the traveller

Overseas travellers are often exposed to infections that they are unlikely to encounter at home. The

Hepatitis A
This is probably the most common vaccinepreventable disease contracted by overseas tra125

Chapter 15 Immunization vellers. Those travelling for a short period in highrisk areas can be protected by passive immunization using human normal immunoglobulin. Vaccination offers good protection and should be offered to those staying in countries where hepatitis A is widespread. It may be worth testing for antibodies in those over 50 years of age or with a history of jaundice prior to immunization. centre is also of great value. Laboratory workers handling infected material should also be vaccinated.

Rabies
This vaccine is usually given combined with passive immunization with rabies-specic immunoglobulin only to people bitten by a rabid animal or by one thought to be infected. It may also be given prophylactically to those with a high occupational risk or who are working in a country in which rabies is endemic.

Meningococcus
The available vaccine offers protection only against Neisseria meningitidis groups A and C, whilst 70% of infections in the UK are due to group B. Countries where groups A and C are endemic and vaccination is recommended include subSaharan Africa, Nepal and northern India. In recent years there have been outbreaks caused by a virulent strain of group W135 associated with pilgrimages to Saudi Arabia, such as the Hajj. Travellers on these pilgrimages should be offered vaccine containing groups A, C, W135 and Y Meningococcus.

Smallpox
With the success of the WHO smallpox eradication programme the vaccine was no longer thought to be necessary. However with the recent threat of bioterrorism many governments have purchased stocks and are considering vaccinating key personnel.

Malaria Yellow fever

This occurs only in parts of Africa and South America. Some countries require an international certicate of vaccination. Avoidance of mosquitoes is the most important protective measure (as with malaria) but immunization with the live virus vaccine obtained from a designated vaccination Each year, some 2000 cases of malaria are reported in the UK in travellers. Most cases arise from failure to take, or poor compliance with, malaria chemoprophylaxis. As yet, there is no effective vaccination against malaria. It is essential for travellers to areas in which the disease is endemic to take appropriate prophylaxis.

126

Chapter 16 Environmental health

Introduction
There has been increasing public concern about the effect that environmental changes might have on the health of the public. This has led to a renewed interest in the real and potential threats from both industrial processes and the pressures arising from urbanization and population growth. Strategies for the control and prevention of diseases caused by noxious physical or biological agents are traditionally based on action directed at containing or eliminating the agent. In the UK this is the responsibility of the Health Protection Agency (HPA). Their website is http://www.hpa.org.uk/. In some circumstances a change in behaviour of the general population is required, for example by encouraging people to use public transport to reduce pollution from trafc. This requires action by other agencies such as local authorities, or the Department of Transport or Education. Adverse effects on health due to environmental conditions can be acute or chronic. An example of an acute effect is an epidemic of respiratory disease brought about by a sudden increase in air pollution, or poisoning due to a chemical spill. Exposure to radiation can have long-term effects, for example cancer or fetal abnormalities. The long-term effects of adverse environmental inuences are often unknown at the time of the exposure. These are thus considered as potential or unproved risks. Public health doctors have a duty to warn of po-

tential as well as known risks. The design of a rational and effective programme to protect against infectious diseases or to reduce the harmful effects of environmental pollutants requires a clear understanding of the relationship between the agent, the environment and humans in each particular instance. Account must be taken of the properties of the agent that affect its ability to cause disease, the ways in which individuals and populations react to it and how the environment can affect the balance between the two directly and indirectly. It is often difcult to communicate the level of risk to the general public. The public wants to know not only the relative risk, but also their absolute risk of disease in order that they may make value judgements about various pollutants and other hazards. Pollution of the environment is increasingly seen as not only producing physical disease, but also having social and psychological consequences. Thus, although doctors are still concerned with agents such as microbes, chemicals and ionizing radiation, which cause physical disease, noise, for example, causing social disruption and psychological stress, is of increasing importance. Global issues such as the destruction of the ozone layer and global warming are also attracting increasing public concern and demand attention.

The social environment

In many respects, highly developed societies pro127

Chapter 16 Environmental health vide a safer environment than those that are less developed. This comes about partly through better environmental sanitation, good housing, clean air and other physical conditions. Moreover, better education and the provision of better personal and preventive health services lead to an awareness of the importance of a healthy lifestyle. However, economic development also involves industrialization and urbanization. The consequences of these go beyond possible damage to the physical environment. They may lead to disruption of old cultures, weakening of family ties and the creation of communities where support for the less competent members has to be provided by welfare services rather than through an integrated community support system. Within any society, the poorest tend to be the least healthy. The consequences of poverty, such as poor standards of nutrition, housing, medical services and education, favour high disease rates. The converse also applies: those who suffer from disease, such as the physically and mentally disabled and those with chronic ailments, have the least earning capacity. Persistent disease in an individual can lead to the phenomenon of downward social class migration. Since the individual is unable to retain the more demanding types of job they may be forced to live in progressively poorer circumstances in which they are exposed to greater environmental hazards and risks of disease. This can give a further downward twist in a cycle of deprivation. Urbanization in general leads to the creation of wealth and in most western countries is reected in the better health of the majority. However, the large populations who come to live close to industrial installations are often exposed to a variety of related health risks. Again, it is the poorest and most disadvantaged who are often forced by circumstance to live in these unhealthy environments. This affects their health and that of their children. Contrary to hopes and expectations, since the inception of the NHS there is little sign that the inequalities in health status between social groups in the UK is decreasing. Indeed, in some cases they may be increasing. The facts were documented in a report, Inequalities in Health (the Black Report), 128 published by HMSO in 1980. The report drew attention to the link between these persistent inequalities and the socioeconomic factors inuencing the material conditions of life of poorer groups, especially children. Its ndings were reviewed, updated and substantially conrmed by Whitehead in The Health Divide, published by the Health Education Council in 1987. Further studies such as the Independent Inquiry into Inequalities in Health Report chaired by Sir Donald Acheson have failed to demonstrate any marked narrowing of the divide. The issue was raised again in the Chief Medical Ofcers report for 2001 available through the DoH website.

Causes of pollution

Pollution
Air pollution Water pollution Sewage and waste disposal Ionizing radiation Industrial accidents

Air pollution
Air pollution in industrial areas arises mainly from combustion of hydrocarbon fuels. The two principal sources are power stations and motor vehicles. A number of pollutants have been identied as causes of ill effects among exposed individuals and populations. These include the following. Sulphur dioxide from the burning of coal or heavy oils. These were the principal sources of the historic London smogs. Suspended particulate matter. This can be identied through ltration methods and is produced by both vehicle exhaust fumes (mainly diesel) and industrial processes. Lead from petrol fumes has been of concern for some years, leading to the wider use of unleaded petrol in some countries and prohibition of leaded fuel in others. Hydrocarbons in the atmosphere from both vehicle exhausts and industrial processes. The poten-

Environmental health Chapter 16 tial carcinogenic action of the complex hydrocarbons that replaced lead in petrol may be a cause for concern. December 1962, London again experienced atmospheric conditions similar to those in 1952 (temperature inversion). The excess number of deaths on this occasion was about 700. Whilst the UK now has few such problems smog is still a cause of ill

Weather conditions
Occasionally, weather conditions arise in which there is temperature inversion, i.e. a warm air blanket covering a layer of cold air at ground level. In cities, this leads to the trapping and rapid accumulation of pollutants known as smog. Such high concentrations of pollutants can cause epidemics of respiratory disease.

1000

Deaths

750 Deaths per day

0.75

500 SO2 250

0.5 SO2 ppm

2 Smoke (mg/m3)

Acute health effects

A dramatic example of the acute effects of air pollution was the infamous smog in London in December 1952 (Fig. 16.1) when it was estimated that the fog was responsible for the deaths of 35004000 people. This led directly to the passing of the Clean Air Act (1956). This empowered local authorities to establish smoke-control areas. As a result, air pollution by smoke declined rapidly in the UK (Fig. 16.2). The benet was seen when, in

0.25

Smoke 1st 8th December 1952 15th

Figure 16.1 Death and pollution levels in the London fog of December 1952. (From Reports of Public Health Medicine Subject 95. HMSO, London, 1954.)

500

400 Emission (1000 tons) 300 Concentration (mg m3) 200

SO2

100

Concentration (mg m3) Smoke

Figure 16.2 Changes in the emission of smoke and sulphur dioxide and their concentrations in London air, 195868.

Emission (1000 tons) 0 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 Year (195868)

129

Chapter 16 Environmental health health in cities such as Shanghai, Los Angeles and Mexico City.

Water pollution
The prevention of water-borne disease rests on the purication and protection of supplies. Adequate and safe water supplies are essential to health. To be safe, drinking water must be free from contamination with both pathogenic microorganisms and harmful chemicals. The most serious infections spread particularly by water are cholera, typhoid and dysentery. These are due to the contamination of water supplies by human excreta. In countries with modern systems of sewage disposal and domestic water supply, spread by this route is extremely rare. Storage assists the purication of water by sedimentation of suspended matter and by biological action. It is further puried by ltration through sand or chemical lters. Finally, it is sterilized by chlorination, which oxidizes organic matter and kills any remaining microorganisms. The dose of chlorine is controlled in order to maintain a small residual amount of free chlorine in the public supply. The water is then distributed through a closed system of pipes and service reservoirs. Its purity is monitored by regular sampling at various points in the distribution system.

Long-term health effects

The long-term damage to health created by air pollution is difcult to separate from the harmful effects of other factors such as tobacco smoking, but acute and chronic chest illnesses are more common in children and in older people living in areas with persistently high levels of pollution. More recently, the contribution of the burning of fossil fuels, especially in power stations, to the phenomenon of acid rain with its destructive effects on the forests of central and northern Europe, has been highlighted. This and the damage to the earths ozone layer caused by the use of chlorouorocarbons as propellants in aerosols and as coolants in refrigerators and freezers have become matters of grave concern to ecologists.

Strategies for control

The Clean Air Act of 1956 has had a major impact in the UK in reducing air pollution from the burning of fossil fuels. Monitoring of the emissions from power stations and industrial factories to ensure they comply with the law is the responsibility of environmental health ofcers employed by local authorities. The strategy to reduce lead in exhaust emissions from vehicles was initially encouraged in the UK by the differential application of duty so that unleaded petrol was less heavily taxed and therefore more attractive to car owners. The alternative strategy to ban leaded fuel is now in place. Despite this, exhaust emissions continue to be a cause for concern. This has led the European Union to require the tting of catalytic converters to all new cars, and vehicles with unacceptable exhaust emissions cannot be licensed. The removal of chlorouorocarbons from the atmosphere is being achieved by a number of voluntary agreements backed by the inuence of powerful environmental groups such as Greenpeace and Friends of the Earth whose activities have encouraged individuals to shun the use of aerosols and refrigerators which contain chlorouorocarbons. 130

Chemical pollution
Chemical pollution of water may arise from the discharge of efuents from factories into rivers and streams and also from the use of pesticides and fertilizers by farmers in water catchment areas. A classic example of industrial pollution of water occurred in Minimata Bay in Japan in the 1950s. In this instance, pollution with mercury led to contamination of sea water which entered the food chain through sh. The result was over 100 deaths in humans, paralysis of many hundreds of others and the deaths of thousands of domestic animals. Generally in the UK, monitoring by the water authorities prevents chemical pollutants reaching a level that is harmful. The protection of water supplies is effected through legislation that prevents individuals and companies from polluting water sources through the discharge of industrial wastes. This has been strengthened by European Union

Environmental health Chapter 16 legislation. The prevention of run-off of nitrates, fertilizers and pesticides from farmland is a problem which may require action. Problems have also arisen in some special circumstances. For instance, the addition of alum to water supplies in order to make the water clearer can lead to problems for people on renal dialysis. This is because the aluminium salts become concentrated and can cause encephalopathy in such patients.

Sewage treatment
In modern sewage treatment plants, after separation of solids by ltering and sedimentation, the liquid sewage is puried by biological oxidation. The nal efuent, which is both clean and safe, is usually discharged into rivers (often to be withdrawn further downstream for water supplies!). Unfortunately, some seaside towns still discharge raw sewage into the sea, sometimes even above low-tide level. This practice leads to offensive pollution of beaches and under certain circumstances may cause a hazard to bathers. Where there is no public sewage disposal system, for example in remote rural areas and on campsites, excreta are disposed of by using chemical toilets or septic tanks.

Fluoridation
Where the natural uoride content of water is high the prevalence of dental caries is substantially less than in low uoride areas. Controlled experiments have shown that this natural benet can be obtained by articial uoridation of water supplies to a level of 1 p.p.m. Maximum protection is achieved when uoridated water is consumed throughout the years of tooth development, and this benet is maintained into adult life. Objections have been raised to the practice of uoridation of public water supplies on the grounds that it is an invasion of individual liberty and that it has potential dangers. However, trials have failed to demonstrate any harm when uoride is added at the recommended levels. Relatively few water authorities uoridate their supplies but the practice is now actively encouraged by the health departments in the UK. Probably the most signicant benet to the population from uoride has been through the use of uoride toothpaste. However for those underprivileged children who are not encouraged to clean their teeth, or whose mothers do not ingest extra uoride during pregnancy, the benet is lost, and without uoridated water supplies there is further disadvantage.

Ionizing radiation
Humans have evolved in an environment bathed in ionizing radiation. Today, most of the ionizing radiation to which a population is exposed still comes from natural sources. Consequently, we are unable to calculate the attributable risk associated with exposure to low levels of ionizing radiation from other sources. However, the ill effects of high doses of exposure are well known. This has led to concerns about the safe levels for both individuals and populations. In addition, the potential risk to the public from nuclear war and industrial and military accidents has led to warnings from concerned physicians. The nuclear accidents at Three Mile Island (USA) in 1963 and Chernobyl (Russia) in 1987, as well as a number of accidents in nuclear powered warships, clearly demonstrated that these fears are well founded. Ionizing radiation can be in the form of X-rays, gamma rays (electromagnetic radiation) or alpha rays and beta rays (particle radiation). Over 85% of the radiation to which people are exposed in the UK comes from natural sources. Around 12% comes from medical sources and around 1% from nuclear fallout and industrial processes. Individuals can be exposed to very different levels of radiation. Some occupational groups such as miners, nuclear industry workers and radiographers/ radiologists may be exposed to much higher 131

Sewage and waste disposal

The provision of an efcient sewage and waste disposal system was probably the single most important public health measure taken in the 19th century. Although this is now taken for granted, it remains central to the protection of food and water supplies, as well as to the maintenance of a clean and safe environment.

Chapter 16 Environmental health amounts of ionizing radiation than the general population. The acute effects of exposure to high doses of radiation include radiation burns, radiation sickness and death. The long-term effects following exposure to high doses have been shown to include cancer (including lung, bone, thyroid and breast cancer) as well as leukaemia, non-Hodgkins lymphoma, congenital abnormalities and thyroid disease. Information about ionizing radiation comes from special events such as by following exposed cohorts from Hiroshima, Nagasaki and Chernobyl, or from people with occupational exposure. In addition, the exposure of large numbers of patients to high dosages of X-rays has provided information about long-term effects. Examples of medical exposure include 40 000 children who in the 1940s had ringworm treated with X-rays to their scalp until their hair fell out, and tuberculosis patients who had large numbers of chest X-rays. Both groups showed an excess risk of death from cancer. Nowadays in the UK, physicians are interested in the effects of ionizing radiation on the general population, on people living near nuclear power installations or weapon factories and on those at risk due to their occupation. A cluster of cases of leukaemia and non-Hodgkins lymphoma around the nuclear power installation at Sellaeld generated particular interest. The cluster has been investigated using both a casecontrol study and a cohort study, but despite the high relative risks for those children living within 5 km of Sellaeld and for children whose fathers worked at Sellaeld a satisfactory causal explanation has not been found. methyl isocyanate gas caused over 2000 deaths and has led to over 500 000 claims for compensation. This was an example of an industrial conglomerate siting a factory close to a residential population in a developing country. Having suffered the horrors of poisoning from the accident, the local population had neither the medical resources to deal with the disaster, nor the legal resources to seek appropriate compensation for the accident. Smallerscale accidents happen frequently around the world and threaten local communities. Prevention in these circumstances not only relies on high standards in the workplace but also depends on sensible planning strategies, which site hazardous industrial processes away from residential populations.

Global health
The concerns of ecologists about the depletion of the ozone layer and acid rain have already been mentioned. In addition, the increasing proportion of carbon dioxide in the atmosphere seems to be leading to an increase in the global temperature, which potentially could cause melting of the polar ice caps and a raising of the oceans levels. This will threaten many island communities. Global warming will also have potential adverse effects on agriculture, which may further exacerbate the nutrition problems of many developing countries causing a deterioration in the health of the world population. Global warming and other global issues were the focus of a 1992 WHO conference in Rio de Janeiro. This led to an acceptance that action is required by all member countries to reduce the use of fossil fuels and to stop deforestation and for joint action to protect the environment. A follow-up Earth Summit in Johannesburg in 2002 reafrmed the need for sustainable development as being a central element of the international agenda. However, the conference was seen by many as a failure, with there being few gains in the 10 years since the Rio de Janeiro Conference.

Industrial accidents
The general public are not only at risk from accidents that lead to nuclear radiation exposure but are also at risk from accidents involving the transport or storage of a wide range of chemicals. The accident at Bhopal, in India, involving the release of

132

Chapter 17 Screening

Introduction
Screening is the practice of investigating apparently healthy individuals with the object of detecting unrecognized disease or its precursors so that measures can be taken that will prevent or delay the development of disease or improve the prognosis. The rationale behind use of the screening to reduce morbidity and mortality is discussed below. In many diseases, the pathological process is established long before the appearance of the symptoms and signs which alert people to the need to seek medical advice. By this time, the disease process and the consequent damage may be irreversible or difcult to treat. For example, in phenylketonuria (an inborn error of metabolism) the abnormality does not usually declare itself before irreversible brain damage has occurred. This can be averted if the condition is detected in the neonatal period and the affected infant is given a diet low in phenylalanine. In other diseases, patients with signs of disease, for example a woman with a lump in the breast or a person with impaired vision, may fail to consult a doctor because the symptoms are not sufciently troublesome or because of fear or stoicism or for other reasons. It seems logical to believe that if potentially serious diseases are diagnosed and treated at an early stage many personal disasters may be averted. If so, a programme aimed at their early detection would be a valuable preventive service.

In other diseases it may be possible to intervene at an even earlier stage in their natural history by treating precursor conditions, thereby reducing the risk that pathology will develop. For example, there is evidence that the risk of stroke can be reduced by controlling blood pressure, and that the risk of a woman developing invasive carcinoma of the uterine cervix is reduced by the detection and treatment of carcinoma in situ. In some circumstances it may be possible to identify individuals who are particularly vulnerable to disease, even though as yet no abnormality exists. Active intervention at this stage may reduce subsequent risk. For example, haemolytic disease of the newborn can be prevented by the administration of anti-D antiserum to the rhesus-negative mother of a rhesus-positive fetus. Screening for genetic abnormalities is an important recent development. The purpose of this screening is to identify people who are apparently normal but at risk for having affected children, i.e. gene carriers. The carrier individuals are then able to make informed reproductive choices. The incidence of diseases such as Huntingtons chorea, fragile X syndrome and cystic brosis may be controlled in this way. Another application of screening is to protect the public health. Some individuals may be infected with an organism and, although they have no symptoms, are capable of transmitting it to others. Such individuals are called healthy carriers. 133

Chapter 17 Screening The detection of the organism in such people will be of no benet to them since they suffer no adverse consequences. However, it is often in the interests of the people with whom they come in contact and the wider community that they should be identied. Ideally once identied they should be treated, but in some circumstances it is not possible to eliminate the organism, for example typhoid carriers. When treatment is not possible, it may be advisable to isolate the affected individuals from situations that may expose others to danger. For example, in an outbreak of methicillin-resistant Staphylococcus aureus wound infections on a surgical unit it would be reasonable to screen all the operating theatre and ward staff in an attempt to identify any healthy carriers. Once identied, such carriers would be taken off clinical duties until such time as they were proven to be clear of infection. Screening has become increasingly recognized as a major tool in improving population health. This has led to the formation of a UK National Screening Committee whose remit is to advise ministers on: the case for implementing new population screening programmes not presently purchased by the NHS within each of the countries in the UK; screening technologies of proven effectiveness but which require controlled and well-managed introduction; and the case for continuing, modifying or withdrawing existing population screening programmes, in particular, programmes inadequately evaluated or of doubtful effectiveness, quality, or value. Their website address is www.nsc.nhs.uk. The use of screening in disease control involves some important assumptions. Some programmes, for example, rest on the assumption that a pathological process can be detected reliably before it is clinically manifest and that, if it is so detected, it can be reversed, arrested, retarded or alleviated more readily than if treatment were delayed until the patient presented with symptoms. For instance, the cervical cytology screening programme depends on two assumptions neither of which has ever been scientically proven. The rst of these is 134 that carcinoma in situ, the condition that the screening process detects, commonly progresses to invasive carcinoma. The second is that invasive cervical carcinoma is invariably preceded by a phase of carcinoma in situ. If either of these assumptions is invalid, the rationale of the programme fails. Moreover, it is impossible, for obvious ethical reasons, to carry out the long-term studies that would be required to test them. Thus, the benets of some screening programmes are theoretical rather than proven, and in future it will be desirable to test the effectiveness of screening programmes with randomized controlled trials before their introduction. Sometimes, the early detection of disease serves only to extend the period of awareness that it is present without improving the prognosis. Furthermore, in any screening programme, cases with a long and relatively benign natural history are more likely to be detected than those with a rapidly progressive and fatal outcome. The dividends from screening in these circumstances can be disappointing, unless the interval between successive examinations is carefully timed to take account of variations in the natural history of the disease in question. Before embarking on any screening programme it is necessary to consider three further important points. Ethics In contrast to clinical practice, which involves the patient asking for the doctors aid to treat established symptoms, in screening programmes apparently healthy people are invited to present themselves for examination. They have the right to assume that this will benet them, or at least will do them no harm. Cost Screening large numbers of people is expensive and can divert both staff and nancial resources from other health services. It is essential therefore to evaluate screening programmes adequately before they are introduced and to weigh the potential dividends both for the individuals screened and for the health of the community against the gains from alternative uses of the same resources, the so-called opportunity cost.

Screening Chapter 17 Effectiveness In order to achieve their aim of reducing levels of morbidity and/or mortality from a particular disease, screening programmes require a high uptake rate, especially amongst particularly vulnerable groups. This is not always easy to achieve as has been found in cervical cytology screening where the most vulnerable groups social classes IV and V have the poorest uptake.

Selective screening
Tests are used to detect a specic disease, or predisposing condition, in people who are known to be at high risk of having, or of developing, the condition.

Single disease screening

Examples Chest X-rays for evidence of pneumoconiosis in coal miners; amniocentesis for detection of chromosomal abnormalities in the fetus in older women; retinopathy in people with diabetes.

Screening programmes
There are two approaches to population screening programmes. One is to restrict screening to members of identiable high-risk groups in a population (selective screening) and the other is to attempt to include everyone regardless of the degree of risk (mass screening). Clearly, it is more economical to focus screening programmes on high-risk groups. Efforts can then be concentrated on securing high participation rates in order to maximize the yield of cases in relation to the effort and expense invested. Whole-population screening is indicated only where it is impossible to dene high-risk groups with sufcient precision to ensure that they include a high proportion of those likely to develop the disease (sensitivity) and the majority not likely to develop the disease is excluded (specicity). Even with so-called mass screening, the programme will normally be restricted to certain broad categories determined, for example, by age, sex, occupation or area of residence. In both selective and mass screening, the programme may be directed to the detection of a specic disease, single disease screening, or include a range of tests for a number of different conditions, multiphasic screening. Types of screening
Selective screening test for disease in high-risk group: single disease screening, e.g. chest X-rays for pneumoconiosis multiphasic screening, e.g. antenatal examinations Mass screening with no reference to risk: single disease screening, e.g. cervical screening multiphasic screening, e.g. biochemical proles on hospital patients Opportunistic screening in general practice

Multiphasic screening
Examples Antenatal examinations; preemployment medical examinations in high-risk occupations.

Mass screening
Large numbers of people are tested for the presence of disease or a predisposing condition without specic reference to their individual risk of having or developing the condition.

Single disease screening

Examples Tests for phenylketonuria and congenital dislocation of hip in infancy; cervical cytology for carcinoma in situ; mammography for breast cancer.

Multiphasic screening
Examples Biochemical proles on hospital patients; routine health check-ups (well-woman clinics, over 75 year olds in general practice, preretirement groups, etc.).

Opportunistic screening
Some screening only occurs when the opportunity arises, for example blood pressure screening for hypertension in general practice, or cervical smears on women using an oral contraceptive. This is of 135

Chapter 17 Screening use because up to 90% of people will see their general practitioner over a 2-year period, so that it is a cost-effective way of reaching a large proportion of the population. before overt disease is apparent. Without knowledge of the full natural history from rst detection by screening to the adverse outcome to be prevented, it is impossible to know what proportion of those screened positive and treated would have progressed to clinical disease.

Criteria for screening programmes

Before the introduction and design of a screening programme, certain criteria should be considered.

Population to be screened
Attention should be paid to the way in which individuals are recruited to a screening programme. Ideally all at-risk individuals should be identied and a systematic effort should be made to screen them all. This may be possible where relevant lists exist. For example, all newborn babies are known and can be screened for phenylketonuria. Those who respond to an open invitation to attend for screening tend to come mainly from self-selected health conscious groups who are often at least risk (low-yield groups) but may also attract those who for one reason or another have delayed seeking advice about existing symptoms (high-yield groups). Frequently, however, it is individuals in highestrisk groups who have the poorest response rates which, unless it can be overcome, limits the potential effectiveness of the programme.

Criteria for screening

The disease The population The test The treatment Severity and frequency, natural history Identication of risk groups, attitudes to screening Sensitivity and specicity of the test, acceptability of the test Effectiveness of early treatment, availability and acceptability of treatment The cost of the programme, screening participation rates

The evaluation

Importance of the disease

Diseases for which a screening programme is proposed should be important in respect of the seriousness of their consequences or their frequency or both. Thus, breast cancer is an important disease because it is both a common cancer and has a high case fatality rate. Successful intervention would be expected to have a signicant impact on mortality and morbidity within a population. Another example is congenital hypothyroidism which is a rare disease but is worth detecting early both because of its serious consequences if untreated and because it is eminently treatable.

Characteristics of the test

No screening programme is possible without a simple, safe and inexpensive test which can reliably discriminate between those who have a high or low risk of disease. The range of normal ndings by the test must be known. It should be quick and easy to use because the object is to test large numbers of people in a minimum time and at a reasonable cost. Unlike clinical practice in which a diagnosis and a decision to adopt a particular treatment is normally based on the history, the ndings from physical examination and the results of laboratory investigations, screening is primarily a sorting process which depends on the results of a single test. This imposes particularly heavy demands on the test. The purpose of screening tests is to divide individuals into two distinct groups: test positive and test negative. However, test positive does not al-

Natural history of the disease

The natural history of the disease must be known in order to identify the points at which the disease is potentially detectable by screening and at which active intervention is likely to be effective: this should be before irreversible damage has been done. Ideally there should be a long latent period 136

Screening Chapter 17 ways mean that the individual has the disease or predisposing condition and conversely test negative does not always mean that they are free from the disease or unlikely to contract it. Conventionally, the characteristics of a test are measured in terms of its sensitivity and specicity (Table 17.1). Sensitivity is the probability that the test will be positive if the disease is truly present: a/(a + c). Specicity is the probability that the test will be negative if the disease is truly absent: d/(b + d). In order to measure the sensitivity and specicity of a screening test, it is desirable to conduct follow-up studies over a period of time amongst people who have been assigned to the positive or negative categories by the test but have not been treated. In some diseases, the presumptive evidence of disease in test-positive individuals is so strong, and the potential consequences of failure to offer prompt treatment are so grave, that it may be unethical to conduct such an investigation. However, if a screening programme is initiated without full knowledge of the test characteristics, problems will arise. Although false negatives will become apparent in due course, these diminish the programmes community benet. Some of the false positives will be identied by subsequent investigations which precede denitive treatment, but those that are not so identied and therefore treated will tend to exaggerate the benets of the programme. They will also waste resources. The problems for patients of being falsely assigned to the positive category are that they may be subjected unnecessarily to time-consuming, unpleasant and potentially harmful further investigations. Occasionally, they may be submitted to unnecessary and harmful treatments, e.g. women in a mammography programme having an unnecessary mastectomy. The false-negative category presents different problems. Clearly, the individuals concerned derive no benet from the test itself. Furthermore, they may be falsely reassured that they are disease free, however carefully the test results are reported to them, and may delay seeking medical aid when symptoms subsequently appear.

Predictive values
Knowing the false-positive and false-negative rates we can ascertain the predictive values of a test: Positive predictive value is the probability of truly having the disease when a screening test is positive: a/a +b Negative predictive value is the probability of being disease free when the screening test is negative: d/c + d

Acceptability of the test

The acceptability of a test is an important factor in the success of a screening programme. Symptomless patients are less amenable to uncomfortable, time-consuming and potentially harmful investigations than those who are seeking medical aid for a problem or potential problem that they themselves recognize.

Effectiveness of early treatment

There is no value in detecting a disease early unless there is an effective treatment that improves the prognosis compared with treatment at a later stage. Consequently, clinical trials of the proposed intervention are required, particularly because the frequency of spontaneous regression in the early stages of disease is often not known. The reversion of an observation in the presumed pathological range to one in the normal range must not be confused with successful treatment. Furthermore, treatments must be assessed in a group that is similar to that which it is proposed to screen. For example, if it is demonstrated that early treatment of mild hypertension reduces morbidity in a group of men aged 4554 years, it cannot be assumed that it will benet men aged 5564 or 6574 years who 137

Table 17.1 Measurement of test sensitivity and specicity. Disease status Present Test positive Test negative Total a c a+c Absent b d b+d Total a+b c+d

Chapter 17 Screening have similar blood pressures, nor that men in the 4554 age group with higher blood pressures will enjoy the same improvement in prognosis. both direct and opportunity costs of a screening programme, must therefore be assessed before its introduction. The calculated cost of a screening programme to the health services should include the costs of all the screening tests performed (both manpower and consumables), the cost of further investigations to discriminate between the true and false positives, the total treatment costs of the positive cases, and the total treatment costs of the false negatives. The benets include the savings on the treatment of cases if they had been allowed to present in the normal way, as well as the social benets related to potentially lost earnings or the loss of a parent and the value of pain and suffering that would have been incurred. These are difcult to quantify. It is of course unreasonable to initiate a screening programme unless there are sufcient resources (trained manpower, hospital beds, technical equipment, etc.) to meet the treatment needs identied by the programme.

Availability and acceptability of treatment

Clearly, there is little point in the early detection of a disease unless the patient is willing to accept and, where appropriate, to sustain treatment at this stage. When a patient has symptoms and believes that medical intervention will bring relief, he or she is more likely to accept the treatment and even endure some side-effects. In offering treatment in the absence of symptoms, the doctor is in a difcult position. Long-term treatment for chronic disorders which cause no obvious and immediate disability, for example hypertension, may not always be successful because of non-compliance. This non-compliance may be because of a misunderstanding on the part of the patient, or because of unacceptable side-effects or forgetfulness. Forgetfulness is probably the greatest problem, as patients have no symptoms to remind them of their condition. Sometimes, delay in seeking medical aid in the presence of symptoms may be because the patient is fearful of the disease itself or of the treatment which he or she thinks may be offered. For example, some women may delay seeking advice about breast lumps because they perceive mastectomy as a more immediate and frightening prospect than the consequences of the disease, or because they see the diagnosis as a deferred but inevitable death sentence. The success of screening programmes for such conditions may also be limited for similar reasons. Termination of pregnancy following antenatal screening presents a stark example of an intervention being absolutely unacceptable to some women. If a woman would not consider termination in any circumstances, screening for fetal abnormality is useless and should not be carried out.

Participation rates
Many screening programmes are only worthwhile if there is a high acceptance rate amongst those invited to participate. Reasons for low uptake can be that the screening test is not acceptable to many people. For example, cervical screening, especially when carried out by a male doctor, will be avoided by some women. This may show up through ethnic or social class variations in the uptake rate of screening. Other inuences on the success of a programme include the level of knowledge concerning the disease being screened for, the manner of the invitation (letters from the persons general practitioner have proved most successful) and the accessibility of the screening venue.

Appropriate intervals for screening

The rst round of screening in a population (the prevalence screen) will have a higher detection rate and be more cost effective than any subsequent or repeat screening (incidence screen). Judging the most appropriate interval for repeat screening requires detailed research.

Cost of screening
Health services increasingly have to recognize that resources of all types are nite. The cost, including 138

Screening Chapter 17 There are two important forms of bias that can be introduced into screening programmes. ethical issues. They were highlighted by the concerns regarding the spread of HIV. It was argued that routine screening of certain groups would help both in understanding the dynamics of the transmission of HIV and in its control. On the other hand, as there was no effective early treatment for HIV infection many believed that the pursuance of such a policy represented an unreasonable and unacceptable intrusion on the privacy of individuals. Once a decision has been made that the public good justies unsolicited invitations for screening, then a number of other ethical issues need to be addressed. People who participate in a screening programme have a right to information concerning the conduct of the programme. They should be aware of the potential disadvantages as well as the expected benets and they should be free to enter or withdraw without coercion. Some programmes can cause unnecessary worry to participants, particularly if they have a positive test. This is sometimes called the labelling effect. In addition, some individuals, including some who are falsely labelled positive, may suffer harm from either the screening test or subsequent treatment. Finally, it is necessary to know whether a specic screening programme is the best way to spend scarce resources. This is a matter of judgement that must be based on good information ideally using a costbenet analysis that takes into account all the costs and benets to both the patient and society. Resources spent on a screening programme may mean that less is available for the provision of health care to others. All of the above ethical questions should be considered by health staff involved in screening programmes whether they be doctors, nurses or managers.

Lead time bias

This is the apparent lengthening of survival achieved by earlier diagnosis rather than by efcacious intervention. Clearly, early treatment will always increase survival time by at least the length of the interval between the presymptomatic diagnosis and symptomatic recognition: the socalled lead time. To demonstrate that an intervention is effective, age-specic death or illness rates must be improved. Increases in survival time can be very misleading when used in isolation as a measure of effectiveness of a screening programme.

Length bias
Interval screening is more likely to identify slowly progressive cases whose prognosis is signicantly better than individuals with aggressive disease. Consequently, cases identied by screening will appear to have a better prognosis than those who have been identied following the appearance of symptoms. In such circumstances, the overall mortality in the population may be unaltered because the screening programme has missed many of the people with aggressive disease.

Ethics
The wider application of screening in the interests of the public health (whether in an attempt to control the spread of disease or in order to understand the pathways by which it is spread) raises difcult

139

Part 3 Health Services

Chapter 18 History and principles

Introduction
Health services fall into two broad categories: personal health services public or environmental health services.

Personal health services

These include the whole range of preventive, treatment and rehabilitative services provided for individuals. In the UK general practice is the main source of primary, domiciliary and ambulatory care. Specialist services tend to be concentrated in general hospitals managed by NHS acute trusts. Long-term care is organized through primary care trusts, local authorities and private and voluntary organizations. the preventive services are mainly provided through primary care trusts and general practice but additional facilities are available in some areas through a variety of agencies.

Public health services

These are concerned with the control and prevention of disease in the community, advice on public policies for health promotion, assessment of the health care needs of the population, and planning and evaluation of health services. The control and preventive function includes the monitoring of disease and the control of factors in the environment that may affect health such as the quality

and safety of air, water and food, the control of occupational and industrial hazards and environmental pollution. In a complex industrial society, health may be affected by public policy in many elds that are not normally thought of as specically health services. For example, education, transport, housing, industrial, commercial and economic policies all directly or indirectly inuence the health and welfare of society. One of the functions of a public health service is to monitor these factors and to provide scientic evidence of their health implications. Public health doctors have traditionally been involved in the development and provision of health promotion strategies with the aim of altering these inuences on health. Most inuence at a national level is exerted by public health specialists in the Department of Health. More recently directors of public health have been appointed to the Government Ofce in each English region. Public health doctors are also involved in health care needs assessment and in providing advice on the provision of health services as well as evaluating their effectiveness and efciency. This role is being developed particularly at the local level within strategic health authorities and primary care trusts.

143

Chapter 18 History and principles volving people outside the family have had to develop. Such involvement made it necessary to create a system of payment for care services and generated the need for professional carers. In modern societies, this is organized and funded by the State. The process whereby personal care evolved from being solely a family obligation to being a professional activity with state involvement in its nancing and supervision was complex. It was inuenced by the structure of societies, changes in the expectations of individuals and developments in medical science and technology. In England, the earliest legislation for the public provision of services for the sick was the Act for the Relief of the Poor (1598), usually referred to as the Poor Law, which required parishes to appoint an overseer of the poor . . . to raise money by local taxation and to provide. . . the necessary relief for the lame, impotent, old, blind and other such being poor and not able to work. This legislation implicitly recognized the relationship between disablement and poverty and it restricted help to those who had no other source of support. In effect, it was a last-resort provision. It was not fully repealed until the passage of the NHS Act in 1946. Early arrangements for the care of the sick were rudimentary. They were provided with shelter, food and basic care. The roles of the doctors and of medicine were limited principally to the care of the wealthy, except through the major charitable hospitals. During the 20th century there was dramatic progress in the development of medical skills and of medical technology. These have affected the shape of medical services in many ways. The new special skills and technologies had to be concentrated in institutions (hospitals) in order that they could develop. This brought about a change in the nature of the hospital from an institution concerned with the general care of the poor to one that was clearly medically orientated. Many people who in previous generations had been looked after at home, turned to hospitals for investigation, care and treatment. Consequently, the social mix of hospital patients changed and they ceased to be the last refuge of the neglected, the destitute and those whose families had dispersed. This changed

Provision of public health medical advice

Department of Health Government Ofces in each region Strategic health authorities Primary care trusts

The history and evolution of both personal and public health services is described in this chapter. In Chapter 19 the present arrangements for the delivery of health services in England and Wales are described.

History of personal health services

A characteristic of human societies is that they accept responsibility for the care of individuals who, through no fault of their own, are unable to care and provide for themselves. In general, these are the elderly, the poor and the disabled. The most basic expression of this obligation used to be care through the extended family, i.e. parents, siblings, children, uncles and aunts, together with others who identify with the family. In most rural societies the main social unit is still the family. Here, families normally live close together and share the same type of work. In these circumstances the personal caring aspect of the familys life is absorbed into its normal activities. In modern industrial societies, the family is no longer always the main social unit. The number of single-person households is increasing. Couples are increasingly choosing not to have children or to delay having a family. People are more mobile, both geographically and socially, and work is normally a separate activity from day-to-day life. For this reason, even though people may appreciate that they have a responsibility to those of its members who are unable to care for themselves, they are often not in a position to assist them. For example, different generations may live in different towns, or daytime jobs may preclude people from devoting sufcient time to the care of an aged relative. Moreover women are increasingly becoming full-time workers and so do not have the time to be unpaid carers as was often the case in the past. Therefore, services in144

History and principles Chapter 18 the standards and nature of care offered within the hospitals. ing the plague in the 17th century, when most physicians left London together with other members of the upper classes. Then, by default, the apothecaries adopted their new role. By the beginning of the 19th century, the apothecaries were well established as doctors to all but the upper classes and were the forerunners of the general practitioner of today. They were not, however, appointed to the honorary staff of the voluntary hospitals.

The medical profession

Doctors
Until the middle of the 19th century there were three types of medical practitioners in the UK: Physicians Surgeons Apothecaries Nearly all of these worked almost entirely outside hospitals.

The GMC
In 1858, physicians, surgeons and apothecaries were placed on a common register maintained by the General Medical Council (GMC) which was charged by law with control over training and qualications. The GMC still regulates the profession and is responsible for setting standards of education, the registration of medical practitioners and dealing with complaints about a doctors tness to practice. Only graduates who have undergone the prescribed training and passed the appropriate exams can be registered. Members of the public who are dissatised with the conduct or performance of any registered medical practitioner may complain to the GMC who will then investigate the complaint and may take action against the doctor, which can include the removal of the doctor from the register.

Physicians
These were university graduates (in England usually from Oxford and Cambridge) who had then qualied for a diploma of the Royal College of Physicians (founded in 1518). Their background was upper class and their practice was mainly among the upper and merchant classes. They attended the voluntary hospitals on a charitable basis.

Surgeons
These originally belonged to the Guild of Barber Surgeons. After 1745 the surgeons split from the barbers and in 1800 formed the College of Surgeons. The College awarded membership to those who passed their exams. Surgeons usually undertook an apprenticeship, attended lectures in anatomy and walked the wards in one of the teaching hospitals before qualifying.

Hospitals
Voluntary hospitals
A few hospitals were established in England by religious orders during the Middle Ages. These include St Bartholomews and St Thomass Hospitals in London. They were founded as practical demonstrations of Christian charity to provide care for the destitute. By 1700, there were fewer than 12 such hospitals in the whole country, most of which were in London. The period during which the greatest number of hospitals was built was in the late 18th and 19th centuries. The voluntary hospitals were supported at rst by church funds, chari145

Apothecaries
The third group, who coexisted with the physicians and surgeons, were the apothecaries. Strictly, they were apprentice-trained tradesmen whose qualication was in making medicines rather than in diagnosis and prescribing. Although they did act as doctors, they were breaking the law in pursuing such activities. They extended their activities dur-

Chapter 18 History and principles table contributions and endowments. They were founded principally as places of asylum and rest for the physically sick and chronically disabled. They were staffed by unpaid doctors (consultants) and, in the teaching hospitals, by doctors in training. Their location was determined in part by local need and in part by the availability of private practice for the honorary staff, which was their only source of income. Outside the main teaching centres, there were other types of voluntary hospital, including cottage hospitals, funded locally and staffed by local general practitioners on a part-time basis. The advent of more sophisticated medical treatments and diagnostic techniques, developed largely in the London teaching hospitals, made the voluntary hospitals become more selective in their admissions. They tended to admit patients who were of good repute, whose stay was likely to be short, and they avoided admitting the chronically sick. The destitute were admitted to the workhouse where rudimentary medical care was provided. By the end of the 19th century, because of the increasing costs of providing a service, they had to introduce a system of payment for those who could afford to pay. The charitable funds were used for those who could not. Most hospitals employed lady almoners whose job it was to establish who should be subsidized and to what extent. Despite the introduction of a semi fee-paying system, the costs of maintaining these hospitals rose faster than their incomes and they became increasingly nancially embarrassed. At the outbreak of the Second World War the Government set up the Emergency Medical Service in order to meet the needs of the large number of military and civilian casualties that were expected. This guaranteed money to the voluntary hospitals to meet the predicted need. After the war, lack of a secure income made a return to their former independent status impossible. Most of them were incorporated into the NHS in 1948. small population units and in order to produce a viable system, groups of parishes combined to administer the Poor Law legislation. These groups were called parish unions. Boards of guardians appointed by the unions were responsible for the day-to-day administration of the institutions. The Poor Law inrmaries were for the destitute sick and were quite unlike hospitals as we know them today. At rst, they did not have any medical staff: nursing care was provided by the non-sick inmates of the workhouse. Over the years, the inrmaries improved, although there was considerable variation in standards. A feature of much of the Poor Law legislation and the legislation governing matters of public health was that, although it gave local authorities discretionary powers to improve the standards and scope of care, it did not place a duty on them to do so. In this lies one of the reasons for the maldistribution of health care resources in the UK. The Poor Law inrmaries were made over to local government authorities in 1929. They then became municipal hospitals. From then until the outbreak of the Second World War a concerted effort was made to improve standards and stafng. In 1939, the municipal hospitals were grouped with the voluntary hospitals in regions as part of the Emergency Medical Service.

Other hospitals
There were two other types of public hospitals during the rst half of the 20th century: fever hospitals and lunatic asylums. The fever hospitals were established to protect the public from infection. Only later were they able to offer treatment. Among them were large numbers of tuberculosis sanitoria. These were built between the two World Wars and are testaments to the high prevalence of that disease and to increasing faith in its treatment. The lunatic asylums had a chequered history. Until 1890, the mentally disturbed were cared for in private mad houses (some with appalling reputations) or in prison or in workhouses (not the workhouse inrmary which was established for the physically sick). The Lunacy Act of 1890 placed a duty on county authorities to provide asylums for those of unsound mind. The London County Council built

Municipal hospitals
The Elizabethan Poor Law enabled parishes to attach inrmary wards to workhouses. Parishes were 146

History and principles Chapter 18 many such hospitals, including nine, with accommodation for several thousand patients, around Epsom in Surrey. The distance from London did not deter the planners as they took it for granted that once patients were admitted there was little chance that they would ever be discharged.

General practice
Specialties developed in the hospitals because of the facilities offered there, while the doctors who worked mainly in the community became known as general practitioners. The majority of the population paid their general practitioner a fee for consultation. In the growing conurbations, the fact that people could be seen free of charge in the outpatient departments of the voluntary hospitals caused some resentment among general practitioners. In order to overcome this, a system was developed whereby patients would only be seen in outpatients if referred by their regular doctor.

Hospitals
Voluntary Municipal Cottage Fever Lunatic asylums

Domiciliary health services

National Health Insurance Act
The Poor Law Commission (1909) demonstrated that a lack of early medical advice often resulted in prolonged sickness and consequent poverty. Its ndings led to the introduction of the National Health Insurance Act in 1911. The important provisions of this Act were as follows.

Domiciliary nursing
At the beginning of the 19th century there were few trained nurses. The need for home nursing was appreciated by the middle of the century and in 1887 the Queens Institute of District Nursing was established. The Institute set and maintained standards of practice and coordinated local voluntary committees.

Mothers and infants

The extremely high maternal and infant mortality in the 19th century led social reformers to look for ways of preventing this waste of life. Important landmarks were as follows. Foundation of the Manchester and Salford Sanitary Association, 1862. This organization employed women to give instruction and guidance to mothers on child rearing. The scheme eventually developed into what is now known as health visiting. The Midwives Act, 1902. This prohibited untrained women from practising midwifery. The Maternity and Child Welfare Act, 1918. This obliged local authorities to provide a medical service for expectant mothers, nursing mothers and children under 5 years of age. The Midwives Act, 1936. This made local authorities responsible for ensuring that there were sufcient midwives to meet the populations need. 147

The National Health Insurance Act (1911)

Free medical treatment from a general practitioner whom the insured person was free to choose (provided the doctor had agreed to participate in the scheme) Doctors who participated in the scheme were paid on a capitation basis, i.e. so much per year per person registered. This was advantageous to the general practitioner as it guaranteed him a regular income for the rst time Weekly payments to insured persons while sick to enable them to maintain minimal living standards

The scheme was restricted to working men whose income was below a specied minimum amount. It did not include retired persons, the wives of working men or their children. The scheme was administered by approved Friendly Societies. In subsequent years, the National Insurance scheme was extended and by 1945 covered the majority of the population.

Chapter 18 History and principles should enable individuals who become ill, or think they have a medical problem, to obtain advice, treatment or sometimes referral to a specialist service. Primary care services are provided in general practice, occupational health services, accident and emergency departments, rst aid rooms and many other places. The precise location of primary care facilities varies from country to country: in most societies there are many alternative sources of such care. In England the primary care trusts have been formed to improve the coordination of the activities of general practitioners, district nurses, health visitors and other community staff.

The NHS Act (1946)

This Act had three major effects: It ensured everyone had free access to a general practitioner It brought the municipal and voluntary hospitals under the control of the (then) Ministry of Health It gave added responsibility to the Medical Ofcer of Health and local authorities in the running of community services

The present tasks of personal medical services

The changes in medical practice during the past 50 years have been revolutionary. Today, access to complex technology, skilled personnel and powerful therapies is taken for granted. Some illnesses that were inaccessible to medical intervention a generation ago, can be treated by methods that are now commonplace. In all branches of medicine, however, there remains a need for the traditional role of the doctor, that of an informed professional carer. In some cases, medicine still has little to offer other than palliation and understanding; in others, once the correct diagnosis has been made the doctors role is simply one of supervising long-term management. Fortunately, most of the population is t and well for most of the time; they only require access to medicine when they become sick. Broadly, the sick can be divided into those who require access to the modern technology of medicine for both the investigation and treatment of their illnesses, and those for whom such facilities are less important than access to carers who have a thorough understanding of them as people and the effects the illness is having upon them. A modern health service must provide facilities, sensitive to individual need, that are accessible to everyone who becomes sick and appropriate caring services for the chronically sick and disabled.

Secondary care
Secondary care is concerned with the provision of specialist services, which are usually provided within hospitals. Medical care that is dependent upon expensive diagnostic and treatment technology is concentrated in hospitals in order to maximize the use of costly equipment and skilled personnel. The task of specialist services is to diagnose, to initiate treatment and, when the equipment to treat is only available at the hospital, to complete the course of treatment.

Intermediate care
Intermediate care has been developed to help reduce the pressure on specialist services by providing more specialized care in the community, to help keep patients from being admitted to hospital and to facilitate early discharge.

Tertiary care
Increasingly, with further sophistication of medical technology, a third or tertiary level of care has evolved. This provides specialty services on a regional or sometimes national basis and usually only accepts referrals from another specialist. Services that are considered tertiary specialties include neonatal intensive care, cardiac surgery, neurosurgery, renal medicine and oncology. These are high-cost services that need to be used efciently and with discrimination.

Primary care
Primary care services are required for the whole population, and for most people they are the rst contact with the organized health services. They 148

History and principles Chapter 18 general practice. In the absence of such resources, many avoidable illnesses will occur to the disadvantage of the individuals and society as a whole. The role of the general practitioner is very broad but few of the activities are exclusive to him or her. The most expensive areas in the provision of medical care are the acute hospitals. It is hard for any society to achieve an ideal balance in its provision of services and there will always be a need to modify provision in the light of the circumstances of each community. In general, rich countries can afford the luxury of expensive technology but poorer and developing countries need to concentrate their sparse resources on personal preventive services, primary care and secondary care that is not dependent on expensive medical technology.

Continuing care
Another type of care is required for the long-term sick and those who do not require the facilities of a high-technology hospital. Ideally, this should be provided as close to the residence of the patient as possible. Sometimes it is feasible to provide it at home. Much of the work of primary care trusts falls into this category and most of this type of care is provided by general practitioners, supported by district nurses, community physiotherapists, health visitors, etc. About 80% of the consultations (whether in the surgery or at the patients home) with general practitioners are generated by about 20% of the population. A large proportion of that 20% are the chronic sick who depend entirely on the general practitioner and his or her primary care team for their medical care. There are other sources of continuing care. Long-term mental health services can be provided in psychiatric units but are increasingly provided by community mental health teams outside the hospital. Similarly institutions for the care of people with learning disabilities have been replaced by community residential accommodation. Other facilities include nursing homes and hospitals for the care of older people, hospices for the dying, homes for the young chronic sick and centres for those disabled by serious permanent injury or disease. In the past, the decision to provide long-term care in a specialist institution rather than in the patients own home was affected more by the social circumstances and the availability of the family and friends to provide basic support than by the patients medical condition. Today, cost is also a consideration and if people can be supported cost effectively in the community by professional carers then this option is increasingly being pursued.

Public health services

Until the early 19th century, there was little public demand for state intervention in matters of health and welfare. During the 1820s and 1830s the socalled sanitary reform movement began to gain momentum. It was particularly promoted by the lawyer and philosopher Jeremy Bentham. He led the push for reform and encouraged the notion that the State should bear some responsibility for the health of its people. His ideas were carried on by his followers of whom the most notable was Edwin Chadwick (Fig. 18.1). Chadwick produced a Report on the Sanitary Condition of the Labouring Population of Great Britain in 1842 which highlighted the economic costs of an unhealthy workforce. This approach gained some support in Parliament and led to the Nuisances Removal Act (1846) which gave local authorities the power to clean up the towns though this was not a requirement on the authority. In 1847, Liverpool appointed the countrys rst Medical Ofcer of Health, Dr W. H. Duncan. The next year, the rst Public Health Act (1848) was passed in the wake of a disastrous outbreak of cholera. This Act which encouraged, but did not compel, local authorities to employ medical ofcers of health also appointed the rst national authority with a responsibility for health in England; the General Board of Health. Opposition led to the disbanding of the Board in 1854. Despite 149

Preventive medicine
Personal health care services must include easy access to preventive medicine (immunization, screening, health education, family planning, etc.). This is provided in a variety of ways including mother and child clinics, school clinics, wellwomen clinics, occupational health centres and

Chapter 18 History and principles then became evident, but the importance of isolation and quarantine, as well as personal and communal hygiene measures, in preventing the transmission of infectious diseases was recognized. The provision of vaccination against smallpox had been a state responsibility since the Vaccination Act (1853) but vaccination against other infectious diseases made little impact until the mid-20th century. When local authorities were established in their modern form in the 19th century, one of their principal roles was to administer environmental health services. Over the years, they have acquired a range of other functions, but their environmental health departments, staffed by environmental health ofcers, continue to be the principal local agencies responsible for monitoring and enforcing many aspects of environmental standards, for example food, water supplies and sewage disposal, air quality, housing and working conditions other than in factories (which are the responsibility of the Health and Safety Executive). They also carry statutory responsibility for the investigation and control of communicable disease in the community, obtaining medical advice for this and other purposes from doctors (consultants in public health medicine) employed by the corresponding health agencies. The Ministry of Health was created in 1919 to exert more effective control over local bodies in the eld of public health. The last important Public Health Act before the National Health Service Act was passed in 1936. It codied and simplied practice relating to environmental and personal hygiene. Thus, by 1946, public environmental health practice and its administration had evolved a structure close to its modern pattern but it remained separate from provision of the treatment of the sick. Recently, the term The New Public Health has come into use (see Chapter 13). Two books were published in the mid-1970s The Medical Nemesis by I. Illich and The Role of Medicine Dream, Mirage or Nemesis by T. McKeown which challenged the importance of high-technology medicine in improving the populations health status. They promoted the view that improvements in nutrition, housing and wealth of the people were the most

Figure 18.1 Edwin Chadwick (180090), author of the Report on the Sanitary Condition of the Labouring Population of Great Britain published in 1842.

its demise, further public health legislation continued to be passed including the 1871 Act (during a major smallpox epidemic), and the Public Health Act (1875) which obliged local authorities to improve provisions for the disposal of sewage, to provide pure water supplies and street cleaning, and to improve housing standards and many other aspects of urban life. The authorities were also then obliged to appoint medical ofcers of health to advise them on matters relating to the health of the community. Interestingly, occupational health services were not included in this legislation and remain outside the NHS to the present day. In the 1870s the discovery of the nature of many infectious diseases began and public health entered the so-called Germ Theory Era in which microbes were recognized as the causes of many of the most signicant diseases of the time. This led to a more scientic approach to the control of infectious diseases. The value of the sanitary reforms 150

History and principles Chapter 18 signicant factors in improving life expectancy and reducing morbidity over the last century. The economic recession caused by the oil crisis of the early 1970s led to a curtailing of spending on health care. This and the escalating costs of modern medicine encouraged a search for alternative ways to improve peoples health. The World Health Organization (WHO) began to expound the concepts of Health for All based on preventive strategies and universal access to basic health services. A WHO meeting held in Southern Russia in 1978 formulated the Alma Ata Declaration on Primary Care. This Declaration proposed a number of strategies for improving health, and emphasized that primary care should be the main focus of national health services in all countries. It was in the context of nancial stringency and growing appreciation of the inuences of environmental and economic and social factors on health that the UK Government commissioned a study led by Sir Douglas Black to examine inequalities in health. The Black Report entitled Social Inequality and Health was published in 1980 and despite initial political resistance became a major inuence on public health doctors thinking about ways to improve health. At the same time, the WHO ofcially adopted Health for All by the Year 2000 as policy: this included a commitment to the idea of equity in health (both within and between countries), a commitment to community consultation and a greater emphasis on prevention and health promotion as strategies to improve health. The Health for All strategy was adopted by the European Region of the WHO which modied and developed appropriate health targets for Europe. These targets included proposed changes to the structure and process of health care as well as detailing targets on specic health outcomes. In the early 1980s, the city of Toronto took the WHO concepts of Health for All and, using community consultation, came up with a plan for a Healthy City. This involved collaboration between the health service agencies and the city authorities; it generated a number of projects that aimed to improve the city environment and peoples health. The European Region of the WHO quickly adopted the idea of Healthy Cities and trials of the strategy of intersectoral collaboration were instigated in a number of European cities. In the UK, Liverpool became one of the rst to embrace this concept and soon over 120 cities worldwide were involved in implementing the Healthy City strategy. In the UK, in pursuit of the Health for All strategy, the DoH in 1991 published a document entitled The Health of the Nation which set out a number of goals for improving public health. This again emphasized a commitment to the pursuit of health, as well as the provision of health care. The strategy involved prioritizing objectives, setting targets and monitoring and reviewing progress. This commitment to public health was conrmed by the incoming Labour Government with its white paper Our Healthier Nation published in 1998.

Milestones in public health in the UK

1842 Edwin Chadwicks Report on the Sanitary Condition of the Labouring Population of Great Britain First Public Health Act and formation of the Board of Health Vaccination Act Germ theory of disease gains momentum Disraelis Public Health Act Ministry of Health formed NHS Act Black Report on Social Inequality and Health Health of the Nation published Our Healthier Nation

1848 1853 1870s 1875 1919 1946 1980 1990 1998

Public health doctors

In 1974, the medical ofcers of health and their personal health service responsibilities were brought into the NHS. At the same time they, together with doctors working in medical administration and in relevant university departments, joined forces to form the new specialty of community medicine. An erosion of the standards in some areas of traditional public health became apparent in the early 1980s, illustrated by a number of serious outbreaks of infectious diseases. This led the Government to set up an enquiry into the public health function. The subsequent report, Public 151

Chapter 18 History and principles Health in England (Acheson, 1988), recommended a return to the old title of public health medicine and doctors specializing in this eld are now called public health physicians. The report redened their role, as outlined below. cial training in the epidemiology of infection, microbiology and clinical infectious diseases. These specialists now work for the Health Protection Agency. They are responsible for surveillance of communicable diseases, investigation of outbreaks and the instigation of control and preventive measures. Research in epidemiology and public health Research in public health involves a multidisciplinary approach, which incorporates statistics, social sciences, health economics and information technology. Specialist areas of research are also developing such as health services research, pharmacoepidemiology and global health, as well as in the more traditional areas concerned with study of the causes and prevention of disease. Most of this work is undertaken in academic departments. Promoting the health of the population This includes the design, management and evaluation of health promotion activities. Health promotion is often initiated at either a national or regional level, and public health physicians with a special interest are usually involved in the identication of issues, the design of appropriate programmes and in arranging their evaluation. Public health physicians are often in the forefront of changes in health services and their roles and responsibilities can change rapidly. For example, in the UK the abolition of the district health authorities and the transfer of many of their responsibilities to primary care trusts has meant that public health physicians have had to rapidly adjust to working in new ways. Whatever the future, it seems certain that a population perspective of health and the expertise of those trained in the relevant specialties will always be essential in a public health service.

Role of public health physicians (1988)

To enquire into all matters which affect the health of communities or population groups To measure health care needs To plan, administer and evaluate services, with particular reference to the prevention of disease To promote health in the community To provide relevant advice to health authorities, central Government and other bodies

In addition, it was recommended that there should be a cadre of public health physicians with special training in communicable disease control. Today, public health physicians have four major areas of responsibility: To advise on the provision of health services These doctors usually work for primary care trusts and are responsible for assessing the health care needs of populations, advising on the purchasing of appropriate services and the evaluation of their effectiveness and efciency. This requires skills in managing, analysing and interpreting information and health statistics. Many of these doctors have areas of special expertise such as the provision of acute or special needs services. At the same time, an understanding of the demographic and social structure of populations and the dynamics of change is extremely important. The control of communicable diseases This is the responsibility of public health physicians with spe-

152

Chapter 19 The National Health Service

Origins
The important report by Sir William Beveridge (Fig. 19.1) on social and allied services, published in 1942, was the culmination of many years of pressure for social reform. It recommended that the State should nance and provide a comprehensive social security system and that it should be underpinned by a comprehensive national health service. It specically proposed that a national health service should provide the facilities which would: . . . ensure that for every citizen there is available whatever medical treatment he requires in whatever form he requires it, domiciliary or institutional, general, specialist or consultant, and will ensure also the provision of dental, ophthalmic and surgical appliances, nursing and midwifery, and rehabilitation after accidents. Apart from humanitarian considerations, which were the principal motivations for the proposals, Beveridge made the apparently logical assertion that such a health service would reduce the costs of social security payments by decreasing the amount of illness in the population. It was thought that this would increase the general appeal of the proposals. When estimating the possible costs of a health service, Beveridge and his colleagues made a further, and, as it turned out, disastrously naive assumption. They assumed that, as the health of the population improved because of the abolition of poverty, better preventive medicine and the elimi-

nation of the long-standing pool of untreated chronic illness, the cost of the proposed National Health Service would fall. They failed to anticipate the possibility of changes in the expectations of the public, the consequences of an ageing population and increases in the costs of medical technology. The wartime coalition government accepted Beveridges proposals for comprehensive national social security and health care systems but was unable to implement them immediately. It charged the Minister of Health for England and Wales and the Secretary of State for Scotland with the responsibility of initiating consultations with representatives of the medical profession, the voluntary hospitals and the local authorities. Discussions began early in 1943 and on 8 February 1944 a White Paper on The National Health Service was published. Its stated objective was: to show what is meant by a comprehensive service and how it ts with what has been done in the past, or is being done in the present, and so help people to look at the matter for themselves. The publication of the White Paper served to crystallize ideas and to stimulate criticism. By the end of 1944 the Minister of Health submitted the suggestions that he had received from all interested parties to the Government. Basically, the proposals involved the Government taking nancial and other responsibilities for the municipal, voluntary and other hospitals, for the general practitioner services as set up 153

Chapter 19 The National Health Service administered by a hospital management committee. The teaching hospitals (both undergraduate and postgraduate) were autonomous from the regional hospital boards. Each had their own board of governors, which worked in close cooperation with the governing body of the associated university institution.

General practitioner services

The administration of general medical services was the responsibility of 134 executive councils. They administered: general medical services (family doctors) general dental services pharmaceutical services ophthalmic services. All of the above services were provided on an independent contractual basis. This means that the general practitioners, dentists, opticians, ophthalmic medical practitioners and pharmacists are not employed by the NHS; they are paid by the NHS for the services they provide. The executive councils had limited disciplinary and planning functions; their main role was that of a paying agency. Technically, general practitioners and other independent contractors were directly accountable to the Minister. In effect, the contractual position of this group of practitioners was little different to that under the 1911 legislation except that the services were now available free to all citizens.

Figure 19.1 Sir William Beveridge, author of the report on Social and Allied Services, published in 1942.

under the 1911 Act, and for municipal public health services and other aspects of personal and preventive medical services. Access to all services was to be without direct charges at the time of use for all residents of the country. In essence, the availability of the then existing services was to be extended, but their basic philosophy and administration changed little. On 19 March 1946, a Bill providing for the establishment of a comprehensive health service was presented to Parliament. Royal Assent was given to the National Health Service Act on 6 November 1946 and the Service was launched in July 1948. The National Health Service, 1948
The administration of the original health service was divided into: Hospital services General practitioner services Local authority services

Local authority services

The local government authorities were responsible for the care and aftercare of patients in the community and for the prevention of disease. Specically, they were responsible for: antenatal care midwifery infant and child welfare district (domiciliary) nursing health visiting school health services immunization ambulance services

Hospital services
Fourteen regional hospital boards were established within which there were 290 hospital groups each 154

The National Health Service Chapter 19 environmental health and a number of other functions relating to the control of infectious disease. control. As a result of a series of enquiries and reports by advisory groups in the 1960s, a major reorganization of the NHS occurred in 1974. The most important aspects of that reorganization were that the country was divided into a number of regional health authorities (DHAs) within each of which there was a number of area health authorities (AHAs), each of which was in turn subdivided into districts (DHAs). The authorities were responsible for the provision of all services other than the independent contractor services within their geographical boundaries. They thus took over many of the responsibilities that had been left with the local authorities in 1948. Where possible, the geographical boundaries of the health authorities were aligned with those of the local government authorities. The independent contractors (general medical practice, general dental practice, pharmaceutical service, ophthalmic services, etc.) became the responsibility of family practitioner committees. The DHAs had a mainly strategic planning and nancial control role, the AHAs planned and managed some of the specialist services whilst the DHAs were responsible for the day-to-day management. In 1975, the Resource Allocation Working Party (RAWP) was appointed to address some of the inconsistencies of funding between regions but not to advise on the total level of funding for the service. Prior to the 1974 reorganization, expenditure per person in some regions was only 55% of that of the richest region. RAWPs main objective was to ensure that there would eventually be equal opportunity of access to health care for people at equal risk. RAWP did not take into account general practitioner services, local authority services or those provided by the private sector. Following RAWP there was substantial redirection of resources to certain less well provided areas. In 1982, the AHAs were abolished. Some of their responsibilities were transferred to the DHAs and others to the DHAs. In the 1980s, there was a renaissance of the philosophy that optimum efciency within an organization was best obtained by exposing the organization to market forces. At the same time, there was a move away from the principle of state ownership. However, the State retained responsibility for politically sensitive areas 155

Early problems
In its early years, the NHS experienced many difculties and shortcomings. The most signicant of these were as follows. There had been a gross underestimation of the cost of the service. The estimated rst-year cost of the NHS was 179 million. It actually cost 400 million. The NHS inherited many old and small hospitals, which had been built under the Poor Law provisions. After the Second World War, the Governments rst priority was to build houses rather than hospitals. As a result there was almost no new hospital building for the rst 20 years of the existence of the NHS. The division of administration of the service between three bodies (hospitals, general practitioners and public health) resulted in lack of coordination and cooperation. For example, many hospitals served several different local authority areas; and all three divisions of the service were involved in maternity services. The NHS had failed to correct the long-standing inequalities in service provision between different parts of the country and between different types of service. The most neglected services were the care of the aged, the mentally ill and people with learning disabilities, together with services for the chronically sick and disabled. The northern regions of the country were poorly provided with hospitals but the areas in and around London had an historical, relatively overgenerous provision. The continued geographical maldistribution of facilities was at least partly due to the fact that there was inadequate capital investment in new hospitals in underprovided regions.

Changes in the 1970s and 1980s

The original tripartite structure of the NHS was seen as a hindrance to the achievement of an integrated and balanced service throughout the country. Equally frustrating was the lack of nancial

Chapter 19 The National Health Service including health and education whilst at the same time introducing the principles of the market place into these services. In the late 1980s, the nancial restrictions placed on institutions by health authorities led some hospitals to seek an alternative funding structure. The idea of independent hospital trusts was born. A bill was passed to allow the creation of NHS trusts, which would be funded directly from the Department of Health (DoH). The move to trust status was to be voluntary and initially most hospitals chose to remain as directly managed units of DHAs. Another key reform enabled general practitioners to hold funds on behalf of their patients including budgets for medicines and some secondary services such as non-urgent surgical and medical services, for example orthopaedics, dermatology, etc. These general practitioners could then choose whether to use their funds to purchase services from the local hospital, from a trust or from the private sector. Choices could be made on the basis of quality and availability of service or on price or a combination of the two. There was an expectation that the introduction of competition into the NHS would control costs and improve quality through the pressure of an articial market. At the same time, the contract between general practitioners and hospitals became more explicit. In many ways these ideas brought about the planned changes. However not all general practitioners took up the challenge of fundholding and after a time there appeared in some areas a two-tier system where patients registered with a fundholding practice received different services than those in nonfundholding practices. In 1997 the new Labour Government reformed the health services yet again, leading in 2002 to the abolishment of RHAs and DHAs which were replaced by 28 Strategic Health Authorities (StHAs) and 309 Primary Care Trusts (PCTs) (Fig. 19.2).

Secretary of State for Health Department of Health

Special Health Authorities (e.g. NICE, HPA, HDA)

Strategic Health Authorities

NHS Trusts

Primary Care Trusts Independent contractors (GPs, dentists, etc.)

Figure 19.2 Organization of health services in England, 2003.

The present management arrangements

The principles that govern the management of the service are that the PCTs have responsibility for purchasing health services for the population of 156

their geographical areas and that, within general strategies and nancial limits, they have considerable autonomy to allow them to respond to local needs. The general strategies and nancial allocations are decided by the Government and the NHS Management Board. The StHAs role is to monitor the PCTs and to ensure that patients are receiving an equitable and high-quality service. The Secretary of State for Health is responsible to Parliament for the NHS and as a member of the Cabinet is able to bring the needs of the Service to the attention of the Government and to argue the case for funds. The Secretary of State is also responsible for the enactment of government policy on health matters, accounting to Parliament for the expenditure of the Service and for its performance. The Secretary of State is responsible to Parliament for: promoting and protecting the health of the nation; providing a national health service in England; social care including oversight of personal social services provided by local authorities in England. The DoH in England provides administrative support to the Secretary of State and the departmental ministers. Its main functions are to assist them by: supplying the information they need regarding the working of the service; advising them on the choices available when making policy decisions; advising on the possible consequences of the available options;

The National Health Service Chapter 19 transmitting policy decisions to the regions; and monitoring progress in their achievement. The Secretaries of State for Scotland, Wales and Northern Ireland have similar responsibilities and are supported by equivalent administrative departments (http://www.doh.gov.uk/). provide general practice services, community services such as district nursing and health visiting (previously provided by community trusts), and public health and health promotion, which were previously provided by DHAs. Many are also responsible for managing cottage hospitals. Perhaps the most important responsibility of PCTs is the purchasing of most of the secondary care services for the people in their area. Secondary care services are the specialist services provided by hospital trusts. In order to purchase services on behalf of their communities PCTs must understand their health needs. Consequently they have developed demographic proles, and local information on mortality and morbidity rates to help them prioritize their services. PCTs are not restricted to purchasing services from local providers. In theory, they can purchase services from whomsoever they wish. Some services may be provided by private practitioners or an alternative acute trust. Indeed some have explored, with government approval, the concept of purchasing health care from other European states. Other services can be provided either by specialists or by primary care. For example, many chronic conditions such as diabetes, care of the elderly or care for people with learning disabilities can be managed either by primary care teams or from hospital-based clinics. Decisions on where to purchase care, and the balance between types of services can vary according to current priorities, perceived quality of care, overall outcomes and cost.

NHS Management Executive

The NHS Executive is concerned with: Regional liaison matters NHS manpower NHS nance NHS support services (building design, maintenance, equipment, etc.)

Strategic Health Authorities

These cover an average population of 1.5 million. The main functions are: to support PCTs and NHS trusts in delivering the NHS plan in their area; in building capacity, i.e. ensuring that there are adequate staff and facilities available; and encouraging improvement in performance by the local health agencies. The chief executive of a strategic health authority is responsible directly to the Department of Health for the performance of the service. He or she advises on planning and strategic matters and is supported by directors of public health, nance, planning, personnel and information technology. The directors of public health of the StHAs are responsible for ensuring that there is a public health network. The network consists of a range of public health practitioners, and as a body they help to ensure that the roles previously carried out by DHAs can continue.

NHS trusts
Most specialist services are provided by NHS trusts which, although part of the NHS, have considerable autonomy within broad guidelines. Each trust has a board of trustees and a chief executive ofcer. There are two main types of trusts. Acute care trusts. These manage large district hospitals (or groups of hospitals) and sometimes regional or tertiary services. Some also manage ambulance services. Mental health trusts. These provide mental health services, including psychiatric inpatient units, community mental health services and in some 157

Primary Care Trusts

PCTs are free-standing, legally established, statutory NHS bodies that are accountable to their strategic health authority. They are organizations that serve a geographically discrete population usually of between 100 000 and 250 000 people. The trusts

Chapter 19 The National Health Service cases forensic services. In some districts, mental health services are run by acute trusts. the one body. These include the NHS public health staff responsible for the control of infectious diseases, the Public Health Laboratory Service, the Centre for Applied Microbiology and Research and the National Focus for Chemical Incidents. Its functions are: to advise the government on public health protection policies and programmmes; to deliver health protection services; to provide information and advice to health professionals; to respond to new threats to public health; and to improve knowledge through research and development and training. Its current website is http://www.hpa.org.uk.

National Institute for Clinical Excellence

The National Institute for Clinical Excellence (NICE) was set up as a Special Health Authority for England and Wales in 1999. It is part of the NHS, and its role is to provide health professionals and the public with authoritative, robust and reliable guidance on current best practice. The guidance covers both individual health technologies and the clinical management of specic conditions. The organization have prioritized their activities to look specically at expensive and new technologies and provide national guidance for health professionals. This is one of the ways that the problem of inequalities can be tackled by ensuring there are national guidelines against which the health services can be monitored. Their website is http://www.nice.org.uk.

Local authorities
Local authorities such as borough and county councils have purchasing powers similar to those of the health authorities, particularly in respect of services for the elderly, some mental health services and services for those with learning disabilities.

Commission for Health Improvement

The Commission for Health Improvement (CHI) was established as an independent body that would review the care provided by the NHS in England and Wales (Scotland has its own regulatory body, the Clinical Standards Board). CHIs stated aim was to address unacceptable variations in NHS patient care by identifying both notable practice and areas where care could be improved. It also published the NHS performance ratings. On 1st April 2004 the Commission for Healthcare Audit and Inspection (CHAI) came into being. It has taken over the responsibilities of CHI plus some of the functions of the Audit Ofce relating to health. The website is www.chai.org.uk.

Special care groups

Care of older people
Elderly patients often reach the point where they are unable to look after themselves at home. This can be precipitated by an acute illness or injury. Support for older patients may involve the provision of a home care worker or admission to sheltered accommodation or a residential home or for those with a serious chronic illness admission to a nursing home may be needed. Long-term care for older people is now a joint responsibility between the health services and the social care department of the local authority. Older people who have suffered an illness and are in need of social support will have both their health and social care needs assessed. The social care departments are responsible for nding appropriate levels of care for patients being discharged from hospital. Patients are then means

Health Protection Agency

The Health Protection Agency (HPA) is a new organization set up to provide an integrated approach to protecting public health. It has brought together a number of different organizations under 158

The National Health Service Chapter 19 tested and may have to pay for the social care component of their care package. The PCT funds their health care needs. If a patients discharge from hospital is delayed after they have been assessed because a suitable care package is not available in the community the social care department has to reimburse the acute trust until the patient leaves hospital. living at home or in accommodation subsidized by the council. Most health districts have one or more special teams that liaise closely with the social services staff, educational authorities and voluntary organizations in order to plan and provide adequate services for this group.

The cost of the NHS

All employed people in the UK pay compulsory weekly or monthly National Insurance contributions, which partly nance the NHS. However, most of the cost of the NHS is met from general taxation. Other nance comes from charges to users, which include dental charges, prescription charges and charges to private patients in NHS hospitals. The level of charges to users and the income they yield varies from time to time. In 2003 the total expenditure on the NHS amounted to 63.5 billion. Most of the expenditure is on hospital services. About 65% of total hospital expenditure is on salaries. This leaves little room for nancial manoeuvre because the numbers of doctors, nurses and other professional staff cannot easily be adjusted to meet short-term changes in need, and cuts in this direction usually lead to a decline in services. The differences in the costs of hospitals of different types are largely due to variations in the numbers of staff needed to provide the services required by various patient groups, for example the clinical staff directly involved in the care of the patient, and the specialist and technical staff who are necessary to enable the clinicians to function adequately (radiologists, pathologists, radiographers, scientists, laboratory technicians, operating theatre staff, intensive care staff, etc.). There are also differences between hospitals of different types in the amount of capital investment required in instruments and machinery. As manpower accounts for the major proportion of hospital costs, the weekly costs are only marginally affected by whether or not a bed is occupied or the appropriateness of its use. Thus, a chronically sick person being cared for in an acute bed costs almost the same as an acutely sick person in the same bed. Extensive misuse of hospital facilities can, if habitual, prove very wasteful. 159

Mental illness
The care of mentally ill people requires the provision of both short- and long-term accommodation and community mental health services. Under the Community Care Act (1990) these are a joint responsibility of the local authority and health. The responsibility for health services is met through the PCT purchasing appropriate services from health care providers. The local authority has a role in purchasing social care in the community such as the provision of accommodation and day care for people with a chronic mental illness.

People with learning disabilities (mental handicap)

About four per thousand of the population have learning disabilities. Under the age of 25 years the majority live at home. In the past, those who were severely disabled or whose families were no longer able to provide total care were looked after by health authorities in long-stay hospitals. The NHS and Community Care Act (1990) transferred to local authorities responsibility for maintaining a register of people with learning disability and for the provision of appropriate accommodation. They can do this either through local authority hostels or, more commonly, by purchasing accommodation and care from private and voluntary organizations. This has meant that the hospitals which traditionally provided care for this group have stopped admissions of new long-stay patients. Consequently, the number of people in private care far outweighs those accommodated in hospital. Much of the cost for care of people with learning disabilities has been shifted from the health sector to local authorities whether they are

Chapter 19 The National Health Service duced health care systems that are funded either by the State or through local authorities. These either underwrite high-risk individuals or offer statesupervised and subsidized insurance. In all of these schemes, the State bears all or part of the cost from general taxation. In most, the user of services pays all or part of the cost and then reclaims a proportion from the insurance fund. This is said to be advantageous because it makes people aware of the true cost of medical care. This type of system operates in most western European countries other than the UK. The British system is unusual in being funded from a combination of direct taxation and National Insurance contributions to which are added a range of charges including prescription and dental charges. There are exemptions from charges for children, the elderly, the unemployed, pregnant women and people with certain chronic illnesses. Medical and other professional practitioners retain a large measure of independence, and responsible use of the service by the public is encouraged by the various charges made to patients. Whatever the source of nance there is a limit to the amount of money that individuals or governments can spend on health. It follows that if a high proportion of the available money is spent on one type of service, for example acute services, less is available for other important aspects of care, for example care of the chronically sick. In a state system, decisions about how the available money should be spent are political, and they must remain so, because the politicians are ultimately accountable for all public expenditure. In privately nanced systems, the balance is determined by the amount of money each individual has and is willing to spend. As the working population has the greatest spending power, this usually results in a growth of acute services to the detriment of services for the elderly, the mentally ill and the disabled.

Planning health services

Objectives
The health service has no single and easily denable objective. Various facilities are provided, including specialist services for the acute sick, preventive services, primary care and care for the chronically sick and disabled. Most of the work involving direct intervention in acute sickness is purely medical in content, i.e. it is mainly dependent upon the technical skills of doctors, supported by other highly trained staff. The care of the chronically sick requires the skill mainly of other professionals, such as nurses, physiotherapists and social workers. Many preventive programmes require action by non-medical professionals, for example teachers and engineers. Health care planning is necessary in order to match needs, demands and available resources within this complex system.

Resources
Financial Staff Facilities

Financial resources
The cost of modern medicine is now such that few people can afford to budget for it out of their income. In most countries where a state-funded system of care does not exist, many people insure themselves against medical expenses. The difculty about this is that the risks of long-term illnesses are difcult for a commercial company to underwrite. Even if this were possible, high premiums would have to be imposed. Chronic ill health affects the individuals earning capacity and its prevalence increases with age. Thus, the most vulnerable members of the community are the least able to maintain payment of premiums. In order to overcome some of the obvious dangers of making each individual responsible for his or her own medical expenses, many countries have intro160

Stafng resources
The second constraint on health service planning is the numbers and types of trained personnel that are available. The principal groups involved are doctors, nurses, therapists and technicians. Ab-

The National Health Service Chapter 19 solute manpower deciencies can arise from a shortfall in national training programmes, from net emigration of personnel, or from a need to increase the staff available to meet rapid advances in diagnostic and therapeutic technology. In the developed world, absolute deciencies are uncommon; manpower problems result mainly from poor distribution. This occurs because: Certain specialties may be less attractive to a young graduate than others. For example, it has always been easier to recruit general surgeons and physicians than it has been to attract people to geriatrics and psychiatry. Usually, people tend to train for specialties that interest them rather than for those that are most needed. Planners have miscalculated and have made insufcient numbers of training places available to cope with future needs. Some areas of the country are more desirable to live in than others; because of this there may be an overprovision in some districts and severe decit in others.

Facilities
The availability of sophisticated equipment can restrict the development of services even if manpower and nance are adequate. This is particularly important for planning the responses to new technological developments.

161

Chapter 20 Health targets

Introduction
During the past 25 years there has been a shift in emphasis from simply providing access to health care to one of attempting to improve the health of the population. This change has been promoted by the World Health Organization (WHO), which in 1975 outlined the concept of Health for All by the Year 2000. Health for All was ofcially adopted as WHO policy in 1981. It advocated that the pursuit of health, rather than only the cure of disease, should be the aim of health policy makers. In addition, it was suggested that policy makers should strive for more equitable health status both within and between countries. The WHO suggested these goals could best be achieved through promotion of healthy lifestyles, the elimination of preventable diseases and the provision of comprehensive health coverage based on primary health care. Health for All suggested that countries should develop health targets that could be monitored to ensure that the strategies of improved health status and equity were being achieved. The European Regional Ofce of the WHO suggested 38 targets to assist member states in setting their own targets. These targets were divided into outcome, process and structural targets. Outcome targets are health targets that aim to reduce the incidence of particular conditions or deaths from certain diseases. Process targets are aimed at reducing the activities that cause disease such as reducing smoking rates 162

or increasing the number of people screened and treated for preventable diseases. Structural targets are aimed at improving the health services in areas where there is good evidence that an intervention will reduce disease such as increasing the number of chest pain clinics or breast screening services.

WHO targets
The 38 targets of Health for All were divided into the following subsets. Outcome targets for improvements in health, for example: eliminating preventable diseases, e.g. measles reducing mortality from diseases, e.g. heart disease and stroke Process targets for activities needed to make these improvements, for example: policies to reduce smoking introducing population-based disease screening Structural targets designed to improve health services management/organization, quality of care, staff training, etc.

Many countries worldwide have adopted the idea of setting national targets aimed at improving the health of the people. The number and types of target have varied widely depending on local needs and the available resources. In July 1992 the UK Government published its strategy for improving health in England as a paper entitled Health of the

Health targets Chapter 20 Nation. Similar documents were produced for the other countries of the UK. In 1998 these targets were updated in the White Paper Our Healthier Nation. The targets related to coronary heart disease, cancer, accidents and mental health were retained but the target related to sexual health was given less prominence. targeted lung, breast, colorectal, testicular, cervical and skin cancer.

Lung cancer
Frequency and trends
Lung cancer is the most common cause of cancer death in the western world. In the UK, there are 38 000 deaths per year. The death rate is higher amongst men than women. It increases with age (Fig. 20.1). The rates of both registrations of new cancers and deaths in women are increasing in contrast to the rates in men, which are decreasing.

Our Healthier Nation (1998)

The key subjects chosen for action were: Cancer Heart disease and stroke Mental health Accidents

Cancers
Cancer still affects almost every family in Britain at some time. Around two in ve people develop cancer during their lifetime, and one in four people die from it. Our Healthier Nation has stated that the aim is to reduce the death rate from cancer in people under 75 years by at least a fth by 2010, potentially saving up to 100 000 lives in total. Cancers account for around 25% of the deaths in the UK. There are many types of cancer and the causes of each differ. Health targets concentrate on the most common and on those whose prevention would add the most years to life. Our Healthier Nation has

Known causes
Ninety per cent of lung cancer deaths are associated with smoking. The differential trends in lung cancer deaths between men and women correlate with their changing patterns of smoking. It is estimated that non-smokers who are regularly exposed to tobacco smoke carry an increased risk of 1030%. The smoking habits of the different social classes is changing, resulting in growing inequity in health status amongst the lower classes due to the effects of cigarette smoking (Fig. 20.2). Other causes of lung cancer include exposure to radon gas, which occurs naturally in some parts of the

6000 5000 Death rate 4000 3000 2000 1000 Figure 20.1 Age-specic death rates for men and women due to malignant neoplasm of the trachea, bronchus and lung (England and Wales, 1998). 0

Male Female

2534 2534

3544

4554 5564 Age (years)

6574

7584

85+

163

Chapter 20 Health targets

Professional Managerial Non-manual skilled Manual skilled Partly skilled Unskilled

Social class

All social classes 0 0 10 20 30 40 50 60 70 80 90 European age-standardized* rate per 100 000 population
* Adjusted for differences in age structure of the population

Figure 20.2 Mortality from lung cancer by social class in men aged 2064 in England and Wales, 19913.

1600 Death rates per 100 000 in 1998 1400 1200 1000 800 600 400 200 0
4 6 9 70 7 4 75 7 9 80 8 4 85 8 9 4 9 4 9 4 5 3 3 4 4 5 9 6

Age

Figure 20.3 Age-specic death rates per 100 000 for carcinoma of female breast (England and Wales, 1998).

30

35

40

45

50

55

UK, and exposure to certain chemicals, e.g. asbestos, arsenic, mustard gas.

60

65

Breast cancer
Frequency and trends

Strategies
Targets to reduce deaths from lung cancer are mainly aimed at reducing smoking. Strategies that have been suggested include advice from general practitioners for patients to give up, wider promotion of the dangers of smoking and the use of adjuvant therapy such as nicotine replacement therapy. There has also been an initiative to improve access to specialist services to try and reduce the case fatality rate by diagnosing and treating patients earlier. 164

Breast cancer is the most common cancer for women in the UK with 41 000 new cases each year. It causes more deaths in women than any other cancer, about 13 000 per year. About a third are in women under the age of 65 years. The death rate increases with age (Fig. 20.3). Breast cancer rates in the USA are higher than those in the UK and the rates in Asian and Hispanic countries are lower.

Health targets Chapter 20 and 16 000 deaths each year. More than 90% of cases occur in people over the age of 50 years. There is a steady increase in the death rate up to the age of 80 years with an excess of deaths in these age groups in men compared with women (Fig. 20.4).

Known causes
Migrant studies have shown that environmental and lifestyle factors are important in the aetiology of breast cancer. Many risk factors for breast cancer have been identied. It is one of the few cancers where the risk is greater in women from a higher social class. Other risk factors include late age at rst pregnancy, late age at menopause and parity (with nulliparous women being at increased risk). Women who have breast fed their babies are at reduced risk.

Known causes
Risk factors for colorectal cancer include a diet high in red meat and fat and low in vegetables and bre. Exercise seems to be protective. People with a history of ulcerative colitis and familial polyps are at increased risk.

Strategies
In the UK, women aged 5064 years are invited to attend breast cancer screening every 3 years. Over one million women are screened annually and 9000 new cancers will be detected. Randomized controlled trials indicate that when a high proportion of the eligible women attend screening, up to a 30% reduction in mortality can be achieved. Consideration is being given to increasing the coverage by including women up to 70 years of age.

Strategies
In March 1998 the Government announced a colorectal screening pilot to assess whether it would be worthwhile introducing a national screening programme for colorectal cancer. Prospective patients are sent a faecal occult blood testing kit and those who test positive are invited to have a colonoscopy. The results of the pilot study are expected shortly.

Colorectal cancer
Frequency and trends
Colorectal cancer is the second most common cause of cancer in the UK with 35 000 new cases

Testicular cancer
Frequency and trends
Each year, there are over 1600 new cases of testicular cancer and it is the most common cancer to

1600 1400 Death rates per 100 000 1200 1000 800 600 400 200 Figure 20.4 Age-specic death rates per 100 000 for men and women due to colorectal cancer (England and Wales, 1998). 0 4044 4549 5054 5559 6064 6569 7074 7579 8084 8589 Age Male Female

165

Chapter 20 Health targets affect men aged 1549 years of age in the UK. It responds particularly well to treatment. Over 9 in 10 patients will be cured and fewer than 100 men a year die from testicular cancer.

Cervical cancer
Frequency and trends
The importance of cervical cancer as a health target relates to the fact that it is the second most common cancer in middle-aged women (after breast cancer), and potentially the outcome can be modied by comprehensive screening and effective treatment of precancerous conditions. There are around 3200 new cases and 1100 deaths each year. The annual number of deaths from cervical cancer has fallen steadily since the 1950s and is now falling by 7% per annum (Fig. 20.5).

Known causes
Risk factors include cryptorchidism (an undescended testicle), having a close relative who has had testicular cancer, infection with the human immunodeciency virus and having Klinefelters syndrome. Testicular cancer is most common in afuent Caucasians and with the exception of New Zealand Maoris, the disease is rare in nonCaucasian populations.

Strategies
A simple, regular self-check can help detect the early signs of the disease. Although most lumps found by self-examination are benign they should be reported to a doctor as soon as possible. Information leaets about testicular cancer entitled A Whole New Ball Game are available from the relevant charities such as Cancer Research UK (website http://www.cancerresearchuk.org/).

Known causes
Cervical cancer is more common in women who become sexually active at a young age, in those with multiple sexual partners and in those whose regular sexual partner has had multiple partners. It has been shown to be associated with the sexual transmission of human papillomavirus, particularly types 16 and 18. There are positive associations with child bearing, with an increased risk for

3000

2500 Number

Number

2000

1500

1000 1953

1963

1973 Year

1983

1993

Figure 20.5 Number of deaths from carcinoma of the uterine cervix in England and Wales, 195398.

166

Health targets Chapter 20 those who start having children at a young age, and the risk increases with increasing parity. There has also been shown to be an association with smoking. Barrier methods of contraception reduce the risk of cervical cancer.

Known causes
Both malignant melanoma and squamous cell carcinoma are associated with excess exposure to ultraviolet radiation and are thus potentially preventable. The increase is thought to be related to exposure to sunshine. The hypothesis that malignant melanoma is associated with exposure to sun is supported by the fact that it is more common in the higher social classes and where Anglo-Saxon populations are resident in tropical regions such as Queensland (Australia) and parts of the USA. Young children who have been sunburnt have an increased risk of malignant melanoma later in life.

Strategies
The death rate has fallen because of better treatment and because of the introduction of the cervical screening programme, which offers screening to all women aged 2064 years of age.

Skin cancer
Frequency and trends
There are three main types of skin cancer: basal cell cancers, squamous cell cancers and malignant melanomas. Although melanoma is a relatively rare skin cancer, it is important because it is the most likely to metastasize and is the most likely to cause premature death. There are around 2000 deaths per year in the UK due to skin cancer, of which 75% are due to malignant melanoma. Death from malignant melanoma is rising by about 5% per annum (Fig. 20.6). Non-melanotic skin cancers affect over 30 000 people a year in the UK and cause about 500 deaths.

Strategies
The Government has spent 10 years raising awareness about ultraviolet light and the risk of skin cancer. It continues to promote avoidance, the use of sun blocks and targeting campaigns at high-risk groups. However, a problem with a primary prevention programme to reduce the incidence of a cancer is that there is likely to be a long lag time between the initiation of the programme and changes in incidence or mortality.

Male 26 Death rates per million 21 Female 16 11 6 Figure 20.6 Age-standardized death rate per million for men and women from malignant melanoma in England and Wales, 19742002. 1 1971 1981 Year 1991 2001

167

Chapter 20 Health targets data usually exclude deaths in people over the age of 75 years. There has been a steady fall in death rates from heart disease in adults in England since the 1970s and similar reductions have been seen in stroke (Fig. 20.8). Despite these reductions, comparisons with countries such as France and Japan (which have much lower recorded death rates from CHD (Fig. 20.9)) encourage the belief that further substantial reductions in the incidence of CHD are possible.

Heart disease and stroke

Frequency
Coronary heart disease (CHD) is due to atheroma of the coronary arteries and is the largest single cause of death amongst men and women in the UK. CHD and stroke together account for 65 000 deaths per year in people under 75 years of age. In those aged under 65 years it accounts for one-third of deaths in men and one-fth in women. The incidence of CHD increases with age, and is greater in men than in women (Fig. 20.7). Often in older age groups, the certied cause of death can be arbitrary. Consequently, comparative mortality

Known causes
The risk of CHD is increased signicantly in relation to three key risk factors: smoking, hyperten-

25 000 20 000 Death rates per million 15 000 10 000 5 000 0 2534 3544 4554 Age 5564 6574 7584 Male Female

Figure 20.7 Age-specic death rates per million for men and women due to coronary heart disease (England and Wales, 1998).

30 Death rate per 100 000 population 25 20 15 10 5 0 1970 1975 1980 Year 1985 1990 1995 Figure 20.8 Death rate per 100 000 from stroke in England, 196996 (3 years average adjusted rates). (Reproduced with permission of the ONS.) Rate of decline slowing down

168

Health targets Chapter 20

Strategies
United States United Kingdom Sweden Japan France Australia 0 100 200 300 400 Mortality per 100 000

Figure 20.9 Age-standardized mortality rates in coronary heart disease in different countries.

4.0 3.5 3.0 2.5 2.0 1.0 0.5 076

Risk 84 91 Diastolic BP 98 105

Figure 20.10 Relation between diastolic blood pressure and stroke from meta-analysis of 420 000 patients followed for 10 years. (MacMahon S. Lancet 1990).

sion and cholesterol levels. Around 30% of the adult population smoke. Studies have shown that the relative risk of death from CHD and stroke is increased in smokers compared to non-smokers and increases with the number of cigarettes smoked. Similarly, there is a positive correlation between blood pressure and the risk of CHD and stroke (Fig. 20.10). There is also a direct relationship between serum cholesterol and CHD. Related risk factors are obesity, lack of exercise and stress. Effective interventions are available for some of these risk factors, either through appropriate therapeutic intervention or through lifestyle changes. Genetic makeup, as indicated by a family history of CHD, and gender are two risk factors that cannot be modied.

The health target set by Our Healthier Nation is to reduce the death rate from heart disease and stroke and related diseases in people under 75 years by at least two-fths by 2010. Intermediate goals have been set by the National Service Framework (NSF) for CHD. These have tended to be less specic and focus on the development of policies based on local need. The NSF has encouraged the health services to develop policies for reducing smoking, promoting healthy eating and physical activity, and for reducing overweight and obesity. They have also encouraged the targeting of high-risk populations such as people originating from south Asia and those from lower social classes to try and reduce health inequalities. Goals for reducing smoking include the introduction of specialist smoking cessation clinics and increasing the number of smokers provided with free nicotine replacement therapy and other adjunctive therapies. Dietary interventions include promoting healthy eating to try and reduce the number of people who are overweight and obese. There are also targets promoting physical activity. General practitioners and primary care teams should identify all people with established cardiovascular disease and offer them comprehensive advice and appropriate treatment to reduce their risks. Medical interventions include advice and treatment to maintain blood pressure below 140/85 mmHg, giving low-dose aspirin (75 mg daily) to high-risk patients, prescribing statins and dietary advice to lower serum cholesterol concentrations, and prescribing warfarin or aspirin for people over 60 years old who also have atrial brillation.

Relative risk

Mental health
Frequency
Deaths attributed to mental illness are principally due to dementia and suicide. There are about 4000 suicides a year, the majority being in people under the age of 65 years. The incidence of suicide has fallen in women since 1980 whilst the rate in men 169

Chapter 20 Health targets has remained relatively static (Fig. 20.11). Mental illness is a major cause of morbidity and utilizes considerable health resources both from primary care and from the specialist mental health services. Around 1% of the population have a major functional psychosis at any one time and one in seven people see their general practitioner in any one year with neurosis (principally anxiety and depression). Prescriptions of antidepressants and anxiolytics are major items within the pharmaceutical budget. There is great scope to improve the efciency and effectiveness of prescribing in this area. Currently, up to 75% of prescriptions for tricyclic antidepressants are at subtherapeutic dosages. Thus, patients are exposed to their side-effects with a reduced likelihood of beneting from the treatment. The key target for mental health is to reduce the death rate from suicide and undetermined injury by at least a fth by 2010 saving up to 4000 lives in total. community mental health services. Other suggested strategies include encouraging general practitioners to recognize and treat a higher proportion of depressed patients, to treat depression with antidepressants at full therapeutic dosages, to elicit an alcohol history from patients and, when appropriate, to take steps to reduce excessive drinking. Many of the issues relevant to improving mental health are outlined in the National Service Framework for Mental Health.

Accidents
In England, accidents result in 10 000 deaths per year and are the most common cause of death in people under 30 years of age. There is substantial variation in the numbers and types of accident with age, sex and social class: for example the annual death rate in males aged 1544 years is four times that of females in this age group, whilst children from poorer backgrounds are more likely to die as the result of an accident than are those from better-off families (Fig. 20.12). The pattern of accidents varies enormously with environmental conditions and personal factors. For example, road accidents occur most frequently in the hours of darkness and in winter months, whereas drownings occur most frequently in the daytime in summer. Alcohol can be a signicant factor in both. Nearly half of all deaths in children are the result of road trafc accidents. Other im-

Strategies
The rst priority in achieving the goals for mental health will be to improve the local and national collection of data and to introduce standardized assessment procedures. It is hoped that the development of comprehensive local services based on local joint planning and purchasing arrangements will ensure continuity of health and social care. This includes plans to ensure 24-hour access to

180 160 Death rates per million 140 120 100 80 60 40 20 0 1940 1944 1948 1952 1956 1960 1964 1968 1972 1976 1980 1984 1988 1992 1996 Figure 20.11 Death rate per million from suicide in men and women in England and Wales, 194098. Female Male

Year

170

Health targets Chapter 20

Professional Social class of father Managerial Non-manual skilled Manual skilled Partly skilled Unskilled 0 10

16.5 15.8 19.1 34.3 37.8 82.9 20 30 40 50 60 70 Death rate per 100 000 80 90 100

Figure 20.12 Mortality from injury and poisoning in children 015 years by social class, 198992. (Roberts I, Power C. BMJ 1996; 313: 7846).

250 200 Death rates per million 150 100 50 0

Male Female

Figure 20.13 Age-specic death rates per million in men and women from transport accidents (1998).

514 1524 2534 3544 4554 5564 6574 7584 Age

84+

portant causes in children include suffocation, burns and scalds, falls and poisoning. There is a sharp peak in the incidence of road accident deaths in males in the 1524 years age group (Fig. 20.13). This peak in road accidents is much less dramatic in females. In people over the age of 55 years, the frequency of accidental deaths increases, particularly as a result of falls in females. Statistics such as these help to identify areas of risk in which there is a special need for preventive effort. The Our Healthier Nation target is to reduce the death rates from accidents by at least one-fth and to reduce the rate of serious injury from accidents by at least onetenth by 2010. Accident prevention is one area which is a crossgovernment concern. It requires concentrated efforts from a number of Government departments

(transport, police, local government, etc.) as well as the DoH in order to achieve substantial change.

Main strategies to help reduce accidental deaths

To reduce the death rate from road accidents To reduce the death rate from accidents in the home To reduce the death rate from accidents in the work place

Road accidents
Every year around 3500 people in the UK are killed in road trafc accidents and 40 000 are seriously injured. Although these gures are better than many 171

Chapter 20 Health targets of our European counterparts there is still room for improvement. The UK Government published a Road Safety Strategy in 2000, which aimed to reduce the number of people killed or seriously injured in road accidents by 40% and to reduce the number of children killed or seriously injured by 50%. The strategy includes improvements in road design, increased driver training and improved vehicle safety. At the same time the wider health issues raised by transport are also being tackled by trying to reduce congestion and by promoting safe walking and cycling. ness, drug abuse and domestic violence can all lead to injury and death. One aspect of home safety that has been highlighted has been re safety. The re service has been encouraged to spend more time on prevention of res and in preventing deaths through the use of smoke alarms. Another strategy area to reduce accidents and deaths in the home is by reducing falls in the elderly. This can be achieved by encouraging safer footware, the tting of handrails and environmental assessments of the home of at-risk elderly patients.

Safety at work Home safety

More people die from accidents in the home than in any other location. Whilst home safety has been improved by better design of domestic appliances (especially electrical and heating appliances), ame-proof childrens clothing and good housing design (especially the protection of stairs and balconies) more could still be achieved. Many home accidents are due to unsafe behaviour rather than an inherently unsafe environment. Thus drunkenAround 67% of all accidents occur in the work place. The targets are to reduce the number of working days lost to work-related injury and ill health by 30% and to reduce death and major injury from accidents at work by 10% by 2010. A third of work place deaths in the UK occur on building sites and the number of deaths in the building industry has been increasing. The responsibility for this lies with the Health and Safety Executive rather than the Department of Health.

172

Chapter 21 Evaluation of health services

Introduction
Over the last 2025 years there have been major changes to the UK health system. These changes have been driven by the desire to improve the quality and efciency of services. In the 1970s there was considerable clinical autonomy and the quality of health care was the responsibility of the clinicians. Following on from then there have been a number of key developments: Principles of general management introduced following the Grifth Report in 1983. Now every NHS trust has a chief executive and a chairperson. Fundholding for general practices introduced in 1990. This was the introduction of the purchaserprovider split where general practices could negotiate and purchase services on behalf of their patients. The market place ensured patients received quality care. Following the NHS White Paper The New NHS; Modern and Dependable in 1997 there was the introduction of the quality framework for health care (Fig. 21.1). Standards of care were made explicit and a complex system of monitoring was introduced to ensure these standards were being applied. Primary care trusts (PCTs) have a responsibility to purchase services. Each PCT has a nite nancial allocation. Because the perceived needs and demands for health care always tend to outstrip resources they have to have clear aims to help them

prioritize the services they purchase. Some health needs will be identied but cannot be met either because the treatments are not available or because they are judged to be too expensive in relation to the expected benet (e.g. some newer drugs for the treatment of multiple sclerosis). Sometimes there is a demand for care but because it is deemed that the treatments are ineffective or unnecessary they are not provided. This is often thought of as a want rather than a health need. Cosmetic surgery or alternative therapies are sometimes placed in this category. It must be remembered that the physical needs of patients are not the only responsibility of the health service; psychological and social needs should also be taken into account. Thus, there can be needs for support, rehabilitation or social services to help maintain and improve health. When thinking about the provision of health services planners will try and balance the health needs with the demands from patients and the supply available in terms of money, staff and resources (Fig. 21.2).

Health needs
Need is determined in part by the nature of the patients problem, and in part by what medical services can offer. Individuals perceive some needs. Other needs are not perceived by individuals but may be recognized by others. Not all people who feel unwell seek professional assistance. They 173

Chapter 21 Evaluation of health services

Patient and public involvement

National Service Frameworks National Institute for Clinical Excellence

Clear standards of service

Professional self regulation

Clinical governance

Lifelong learning

Dependable local delivery

National Performance Framework Commission for Health Improvement

Monitored standards Figure 21.1 The quality framework used by the NHS.

National Patient and User Survey

Needs assessment
When estimating the need for health services, it is useful to look rst at the prevalence and incidence of diseases within the population. This coupled with the demographic data is the minimal baseline information required to estimate need. Interpretation of epidemiological data on need should take into account factors such as age, gender and ethnicity. It is necessary also to take into account whether an effective intervention exists, and the availability of the necessary facilities and resources to meet identied needs.

Needs

Demand

Supply

Unperceived needs
Figure 21.2 A schematic approach to needs assessment.

take action themselves, for example by going to bed for 2 days because of inuenza, or take advice from a friend or relative. Once they decide that they require medical intervention they make a demand on the health service. The doctor who sees the patient may or may not accept that the problem will benet from his or her skills. The only type of need that can be measured without special study is that which creates a demand on the service. 174

An individual who is aware of his or her need for medical intervention has symptoms or signs that he or she associates with illness. However, the professional worker may detect signs of disease that are amenable to treatment in the absence of such symptoms. This is sometimes incidental to examination for another reason or may come to light from screening or health examination surveys.

Demands
The work load of a health service is affected by the incidence of acute diseases and the prevalence of chronic diseases for which care may be required

Evaluation of health services Chapter 21 over a long period. Demand is measured either by monitoring the workload of the service or by special surveys. Demands on services are not always an accurate proxy for need. They are affected by: knowledge of the existence of services; local availability of services; and sectional pressure for intervention. Without knowledge of the existence of a facility or treatment an individual will not make a demand for that service. For example. a patient with a painful hip who is not aware of the benets of a total hip replacement may perceive a need to have the pain relieved, yet take no action. The publicity given to a particular service, for example by a television documentary, inevitably increases the perceived need and therefore increases demand. Similarly, demand is likely to be greater where a particular specialist facility is available locally and this is known to the local population and their doctors. Demand for health services can also be thought of in terms of health need and supply. Many demands are for services where there is also need, such as for emergency care or maternity services. In other cases the demands may be present but the needs are less obvious, e.g. for tonsillectomy in young children, circumcision of male infants or breast augmentation in women. Finally there may be demand and need but no supply. In some developing countries facilities such as renal dialysis or coronary artery bypass grafting are simply not available. In other circumstances there may be a need but the technology does not exist. However, a scientic breakthrough can rapidly change this, creating new demands on services. The development of sildenal for the treatment of erectile dysfunction might be an example. continual improvement has been adopted. This requires the structures and processes involved in health care to be continually modied and careful monitoring will verify the improvement in outcomes. This entails the application of the so-called quality cycle in which a standard of care is set, the process of care given is monitored and the outcome is measured, then new standards are adopted. Measurement of quality can involve every patient treated, as is applied, for example, to renal dialysis patients and those undergoing chemotherapy, or it can involve a sample of patients.

Clinical governance and audit

Clinical governance was a new initiative proposed in the 1997 White Paper. It is the way in which NHS organizations quality assures their services. It involves putting in place the information, methods and systems to ensure good quality care is being provided. It has been described as the framework through which NHS organizations are accountable for continually improving the quality of their services and safeguarding high standards of care by creating an environment in which excellence will ourish. This includes action to ensure that risks are avoided, adverse events are rapidly detected, openly investigated and the lessons learned. Good practice should be rapidly disseminated and systems ought to be in place to ensure continuous improvements in clinical care. Although given a new name, much of the work carried out in the name of clinical governance is audit. Audit involves a cycle (Fig. 21.3) where rstly the standard of care to be achieved is agreed and the services are then measured against this standard. Suggestions are then made as to how this care can be improved, changes are made and a further audit is carried out to ensure the predicted changes have been achieved. Clinical audit examines the total package of care offered to patients. This may involve assessment of the structure and process of care as well as outcomes. It may review not only medical care, but also nursing care, the physical environment and the organization and management of services. Health care can be measured in terms of seven key parameters. 175

Quality in health care

Quality is a nebulous concept. It is a function of both the service provided and the expectation of the customer. Thus, as expectations rise, patients perception of the quality of care they are receiving is likely to fall. The maintenance of quality has often been focused on the elimination of bad or unacceptable practice. Increasingly, the concept of

Chapter 21 Evaluation of health services ment or differences in their outcomes to ensure that services are fairly distributed. Thus, people from minority groups or those of low socioeconomic status, despite a similar or often higher prevalence of disease, may have lower rates of treatment. Even if they are treated on an equitable basis, outcomes may still be worse. The equity of a service can only be judged if these factors are identied and monitored. They are often only remedied by targeting of services to the disadvantaged group.

Repeat the audit cycle The Audit Cycle Implement changes

Set standards

Audit practice

Compare results of audit with standards

Figure 21.3 The audit cycle.

Access
Aspects of care
Efcacy: does it work? Effectiveness: how well does it work? Efciency: is this the best way of doing it? Equity: is it fair? Accessibility: can everyone use the services? Acceptability: is it what they want? Appropriateness: is it what they need?

Access involves the assessment of barriers to care in order to ensure that people obtain the treatment they need when they need it. Barriers can include cost, waiting lists, location of the service or the need to convince a general practitioner of need. Often these barriers are only identied by asking patients directly.

Acceptability Efcacy
Efcacy is the measure of the capacity of an intervention to produce a desired effect. Some services may not be used because of the way they are provided. Issues such as privacy, the gender or attitude of staff, and the setting of the service can inuence the utilization of health services. These factors are often only discovered through patient questionnaires.

Effectiveness
Effectiveness involves assessing clinical outcomes of health care such as mortality rates and survival times. A treatment must show an improvement in clinical outcome, ideally through use of randomized controlled trials, in order to be considered effective.

Appropriateness
Any assessment of health services must measure whether the needs of the population are being met. This requires constant assessment of need and audit of the structure of health services as well as monitoring such indicators as waiting lists.

Efciency
Efciency involves the assessment of the costs of services. The most efcient service will produce the desired outcome at the lowest cost.

Accreditation
In some countries, the concept of accrediting organizations that meet certain quality standards is being adopted within the health service. Accreditation is common practice, for example in the food industry, and is sometimes applied to hospital laboratories. The assessment includes standards of

Equity
Equity involves assessing differences in the needs of those receiving care, differences in their treat176

Evaluation of health services Chapter 21 practice (including the training of staff), adherence to protocols, validity and reliability of diagnostic testing, safety standards, etc. The concept of accreditation is applied to the whole range of hospital and community services in the USA and Australasia: accreditation provides purchasers with an assured quality standard which is taken into account when negotiating contracts. In many cases, accredited hospitals are rewarded by being paid a higher rate for the services they provide. UK purchasers use similar processes to monitor the quality of services provided by contractors. outcome. Examples of outcomes that can be compared include the cost per patient successfully treated or cost per life saved. The disadvantage of a cost-effectiveness analysis is that it may not reveal other positive or negative effects of compared treatments other than those that have been recorded and which are the subject of the analysis.

Costbenet analysis
Costbenet analysis compares two or more treatments or services, by placing a value (usually monetary) on all the accrued costs and on all the benets. Thus, when considering a treatment, the benets may include added years of life. This is assigned a monetary value often based on future potential earnings of the individual. It may also include the costs of continuing care or treatment. This form of economic analysis allows purchasers to compare many different treatments to help them decide which is the best buy. The disadvantage is that it tends to discriminate against the elderly and those with a low earning potential such as the physically disabled and those with learning disabilities because the benets are not easily measurable in monetary terms.

Health economics
The cost of health services has been one of the Governments primary concerns since the beginning of the NHS. The demand for health services can be almost innite but resources are always limited. In this circumstance choices have to be made. Health economics can help decide which are the best choices. Often in the past management of health care has focused on the principle of cost containment. This has the disadvantage that containment may actually reduce efciency. Attempts to save money by undertaking fewer hip replacements does not reduce the xed cost of the theatres, or the need to pay the nurses and surgeons. So whilst overall costs may reduce, the cost per hip replacement will increase. Economists look at ways of better using the money that is allocated to the health services. They use the term opportunity cost in a cash-limited health service, money spent on one aspect of care means that another service cannot then be funded. Health economists also use several methods of comparing costs and outcomes.

Costutility analysis
Costutility analysis accounts for all the costs of comparable treatments, but measures the benets in a common unit (other than money). One common unit that has been used is the qualityadjusted life year (QALY). A QALY combines the quantity of life gained with an adjustment for quality of life. This allows direct comparison between treatments but is less discriminatory against those with limited earning capacity than costbenet analysis. Other measures of the benets of medical care have been developed; for example, in 1993 the World Bank adopted a new and more sophisticated unit, disability-adjusted life years (DALY).

Cost effectiveness
Cost effectiveness measures the cost of one or more treatments or services in comparison to a single

177

Appendices Further Reading and Useful Websites

Appendix 1 Suggested further reading

Armitage P, Berry G. Statistical Methods in Medical Research. Oxford: Blackwell Science, 2001. Ashton J, Seymour H. The New Public Health. Open University Press, 1990. Beaglehole R, Bonita R, Kjellstrom T. Basic Epidemiology. World Health Organization, 1993. Benenson AS. Control of Communicable Disease in Man. American Public Health Association, 1997. Bland M. An Introduction to Medical Statistics. Oxford: Oxford University Press, 2000. Detels R, McEwen J, Beaglehole R, Tanaka H. Oxford Textbook of Public Health. Oxford University Press, 2002. Donaldson RJ, Donaldson LJ. Essential Public Health Medicine, 2nd edn. Petroc Press, 2000. Drummond MF, Maynard A. Purchasing and Providing Cost-Effective Health Care. Churchill Livingstone, 1997. Greenhalgh T. How to Read a Paper. BMJ Publications, 2001.

Joint Committee on Vaccination and Immunisation. Immunisation Against Infectious Diseases. London: HMSO, 1996. McKeown T. The Role of Medicine. Oxford: Basil Blackwell, 1980. Naidoo J, Wills J Health Promotion Foundations for Practice, UK. Second edition. Bailliere Tindall, 2000. Pereira-Maxwell F. AZ of Medical Statistics. Oxford University Press, 1998. Rose G. The Strategy of Preventive Medicine. Oxford Medical Publications, 1992. Sackett DL, Straus S, Richardson S, Rosenberg W, Haynes RB. Evidence-Based Medicine. How to Practice and Teach EBM, 2nd edn. London: Churchill Livingstone, 2000. Scambler G, ed. Sociology as Applied to Medicine. London: W.B. Saunders, 1997. Townsend P, Davidson N. Inequalities in Health (the Black Report). London: Penguin Books, 1982.

181

Appendix 2 Useful websites

Association of Public Health Observatories Bandolier Cancer Registration Cancer Research UK Commission for Healthcare Audit and Inspection Department of Health in England Department of Statistics Health Development Agency Health Protection Agency Hospital episodes statistics Immunization in the UK National Institute of Clinical Excellence Scientic Advisory Committee on Nutrition UK National Screening Committee World Health Organization Healthy Cities

www.pho.org.uk www.jr2.ox.ac.uk/bandolier www.ociu.org.uk www.cancerresearchuk.org www.chai.org.uk www.doh.gov.uk www.statistics.gov.uk www.hda-online.org.uk www.hpa.org.uk www.doh.gov.uk/hes www.immunization.org.uk www.nice.org.uk www.sacn.gov.uk www.nsc.nhs.uk www.who.dk/healthy-cities

182

Index

A abortion 578 induced (terminations) 80, 138 following antenatal screening 138 spontaneous 7980 Abortion Act (1967) 57 Abortion Act (1990) 57 acceptability of service 176 accessibility of service 176 screening test 138 treatment following screening 138 accidents 1702 industrial see industry prevention 1702 accreditation 1767 accuracy of test 31 acid rain 130 Action on Smoking and Health 99 acute care trusts 157 adjuvants, vaccine 116 aetiology see cause age 21 death rates related to 678, 734 fertility rates related to 767 standardization of rates for 678 structure of population 71 susceptibility to infection and 104 AIDS/HIV virus 9, 104 notications 60 screening 139 air pollution 12830 alcohol abuse, health promotion 99100 Alma Ata Declaration on Primary Care 151 anaphylaxis after vaccination 117 animals as models of disease 8 as reservoirs of infection 104, 110 antenatal screening, termination following 138 antibodies passive immunization with 11415 to vaccines 11415 antidepressants 170 antigenic stability of pathogen 106 apothecaries 145 appropriateness of service 176 area health authorities 155 asbestos 9

association chance, distinguishing causes and determinants from 810 strength of 9 Association of Public Health Observatories 182 attributable risk, in cohort studies, calculation 11 audit 1756 autonomy, personal, rights to 102 B Bandolier 182 barrier contraceptives and cervical cancer 423 BCG vaccination 15, 114, 116, 1234 benecence 102 Bentham, Jeremy 149 Beveridge Report 1534 bias 84 avoidance 2930 in cohort studies 33 in denominators 2930 in numerator data 289, 66 in sampling 278 in screening programmes 139 systematic 30 births control methods see contraception information 53 rates 64, 767 illegitimate 79 seasonality and mental illness 18 stillbirths 22, 53, 567, 64 Births and Deaths Registration Act (1968) 53 Black Report 5, 128, 151 blinding 85 blood pressure, elevated see hypertension body mass index 634 Bordetella pertussis see pertussis (whooping cough) immunization breast cancer 136, 1645 casecontrol study case selection 38 confounding variables 41 uoroscopy and 36 oral contraceptives and 10 British Heart Foundation 98 British Nutrition Foundation 100 bronchial carcinoma mortalities 16 smoking and 8, 412 see also lung cancer buildings, hospital, expenditure 159

183

Index
C Canada Lalonde Report 96 Ottawa Charter 97 Toronto, Healthy City strategy 151 cancer 1637 breast see breast cancer cervical 423, 135, 1667 colorectal 165 haematological, radiation and 1312 lung see lung cancer registration 61 skin 167 stomach 23 testicular 1656 Cancer Registration 182 Cancer Research UK 182 Cancer UK 98 carcinoma see specic site cardiovascular disease oral contraceptives and 78 social class and 367 see also heart disease casecontrol studies 1213, 3844 advantages/disadvantages 42 controls, selection 3940 examples 424 risk calculation 401 case fatality rates 64, 65 case(s) difculties and problems in ascertainment/identication 66 selection 389 cause 710 chance association and, distinguishing 810 descriptive studies in determining 1415 of infectious disease 1034 cause-specic rates 64 censuses 512 cervical cancer 423, 135, 1667 Chadwick, Edwin 149 chance association, distinguishing causes and determinants from 810 chemical pollution of water 1301 children accidents 1701, 172 births see births deaths see infants; perinatal deaths immunization see immunization chlorouorocarbons 130 cholera 10, 20, 93 deaths 75 vaccine 125 chronic disease care 149 prevention 923 Cities, Healthy 967, 151 Clean Air Act (1956) 129 climate and infection 106 clinical governance 1756 clinical trials see intervention studies (clinical trials) Clostridium tetani 119 cluster allocation 46 cluster sample 27 Cochrane Collaboration 86 cohort effect 21 cohort studies 12, 327 advantages 33 disadvantages 334 examples 347 risk calculation 401 colorectal cancer 165 Commission for Health Improvement 158, 182 communicable disease see infectious disease community, continuing care in 149 community action, strengthening 97 Community Care Act (1990) 159 component vaccines 116 conception see births; contraception; fertility (rates) confounding variables adjusting for 41 denition 78 congenital malformations 601 congenital rubella syndrome 122 consent for vaccination 118 Consultant in Communicable Disease Control 59, 113 contraception 423, 778, 167 barrier, cervical cancer and 424 education 101 efciency of various methods 78 oral see oral contraceptives control (in trials) and control groups in casecontrol studies, selection/recruitment 3940, 42 in clinical trials, allocation to 46 coronary heart disease see heart disease, ischaemic (coronary) coroners 55 Corynebacterium diphtheriae 119 costbenet analysis 177 cost-effectiveness 177 costs, economic 177 NHS 159 screening 134, 138 costutility analysis 177 cot death 22 critical appraisal, evidence-based medicine 834 systematic reviews 856 cross-sectional surveys 25 crude rates 67 cumulative incidence 65 cycle of deprivation 128 D data 5162 analysis descriptive studies 12 outbreaks of disease 11011

184

Index
capture procedure 61 grouping 634 numerator, errors and bias in 289, 66 presentation, inconsistency 25 routinely collected, problems with 25 web sources 51 see also health information; records databases, general practice 59 deaths see mortalities (rates) demands on services 1745 demography, medical 6981 fertility see fertility (rates) fetal loss and infant mortality 7981 mortality, reasons for decline 745 populations and growth rates 702 transition 724 denominator error 66 Department of Health, responsibilities 1567 Department of Statistics 182 deprivation, cycle of 128 derived infection 107 descriptive studies 1423 data analysis 1523 use of 1415 determinants 7 chance association and, distinguishing 810 descriptive studies in discovering 12 developing countries, demographics 724 diabetes 25 prevalence 26 diagnosis/detection 5 criteria used, variations 245 diet heart disease and 1689 promoting healthy 100 see also nutrition diphtheria 119 diphtheria/tetanus/pertussis (DTP) vaccination 117, 11820 direct standardization 68 discreet quantitative variables 63 discrimination 301 disease/ill-health cause see cause distribution 9 indices of 638 prevention intervention strategies 945 principles 934 see also specic diseases divorce and fertility 77 doctors/physicians history of profession 145 public health 1512 see also general practitioners Doll, Richard 34 domiciliary health services 147, 172 double-blind trial 46 E Earth Summit 132 education see health education effectiveness cost-effectiveness 177 screening 135 efcacy denition 176 vaccines 11718 efciency of NHS 176 elderly accidents 172 care, local authority responsibilities 1589 enteric infections see gastrointestinal infections environment health services concerned with 12732 infection and 106 workplace safety 172 Environmental Health Ofcers 113, 150 epidemics 1819, 10711 common source 108 herd immunity affecting periodicity 105 investigation 10911 propagated 1089 types 1089 epidemiological studies see studies/surveys equity 176 errors (in surveys) 2831 assessment 301 avoidance 29 in health information 657 random 30 systematic 30 Escherichia coli 157 food poisoning 108, 112 ethics clinical trials 47 health promotion 1012 screening programmes 134, 139 ethnicity 22 evidence-based medicine 827 challenges 867 critical appraisal 834 systematic reviews 856 randomized controlled trials 845 search strategies 823 strengths and weaknesses 87 exercise 1001 expenditure on NHS 159 exposure (to agents or experience) radiation leukaemia and 36, 44, 132 lymphoma and 44 special, groups with 323 F family planning see contraception fertility (rates) 769 factors affecting 768 patterns, changes in 789 total-period 79

185

Index
fetal loss see abortion nancial resources see funds Five a Day programme 100 uoridation, water 131 uoroscopy and breast cancer 36 folic acid supplementation, Medical Research Council Vitamin Study 50 Food and Drugs Act (1955) 113 Food in Schools programme 100 food poisoning incidence 17 investigation 10913 Foundation of the Manchester and Salford Sanitary Association 147 Friends of the Earth 98 funds 160 general practitioners holding 156 Primary Care Trusts and 157 sources 159 G gastrointestinal infections mortalities 75 seasonality 17 gender 212 susceptibility to infection and 1045 General Household Survey 27 General Medical Council 145 general practice databases 59 General Practice Research Database 59 general practitioners fund-holding 156 history 147 services 154 see also doctors/physicians generation time 107 genetic inheritance 22 genetic screening 133 genetic susceptibility to infection 1045 genital infections (and sexually transmitted disease) 101 cervical cancer and 1667 notication of episodes 60 transmission 104 geographical factors in epidemiology 1920, 109 german measles see rubella Germ Theory Era 150 global warming 132 greenhouse effect 132 Greenpeace 98 grey literature 86 group(s) (of individuals) with special exposures 323 with special personal characteristics 32 test and control, allocation 46 H Haemophilus inuenzae type b vaccination 11819, 120 Hawthorne effect 28 health denition 51 economics see costs education see health education ill see disease/ill-health indices 638 inequalities 1278, 151 targets 16272 health authorities 15361 local 154, 1589 Strategic 157 Health Development Agency 98, 182 health education 96102 alcohol abuse 99100 contraception 778 denition 96 diet 100 exercise benets 1001 see also health information; health promotion Health Education Council 97 Health for All by the Year 2000 97, 151, 162 targets 124 health information 5162 errors 657 retrieval 823 systematic reviews 856 systems 612 see also data; health education; health promotion; records health needs 1734 assessment 174 Health of the Nation Programme 100, 124, 151, 162 health promotion 96102, 152 denition 96 ethics 1012 new public health 968 programmes 98101 sexual health 101 see also health education; health information Health Protection Agency 127, 158, 182 health services 14352 authorities see health authorities domiciliary 147 evaluation 1737 history and principles 14352 personal 143, 1445 planning 1601 public 1434 reorientating 978 see also National Health Service health workers 1601 Healthy Cities 967, 151 Healthy Schools Programme 101 heart disease, ischaemic (coronary) 1689 mortalities 678 social class and 367 see also cardiovascular disease hepatitis A vaccine 1256

186

Index
hepatitis B vaccine 124 herd immunity 105 HIV infection see AIDS/HIV virus home safety 147, 172 Hospital episodes statistics 182 hospitals 1457 episode statistics 59 expenditure 159 municipal 146 patients in, as controls in casecontrol studies 39 services 154 trusts 1578 voluntary 1456 host (human), in infectious disease as reservoirs/carriers 1034 susceptibility 1045 see also individual(s) housing, infection and 106 humans see host (human); individual(s) hydrocarbons, polluting 1289 hypertension 25, 64, 1378 Medical Research Council trial in treatment of mild 48 I illegitimate birth rates 79 illness see disease/ill-health immunity to infection 105 immunization 11426 active see vaccination passive 11415 routine 11824 schedules 118 targets 124 Immunization in the UK 182 immunogenicity 106 immunoglobulins, passive immunization with 11415 incidence (rates) 645 cumulative 65 incubation period 107 denition 107 Independent Inquiry into Inequalities in Health Report 128 indices of health and disease 638 indirect standardization 678 individual(s) allocation of, in clinical trials 46 autonomy, rights to 102 characteristics 203 preventive programmes and 95 records see records industry occupations in 60 accidents 60, 132, 172 diseases 60 risk to public of accidents 172 see also occupation inequalities in health 1278, 151 infants death 567, 7981 sudden 22, 57 health service development 147 infectious disease 10313 antigenic stability 106 control and prevention 915, 10313 see also immunization death rates from see mortalities (rates) enteric see gastrointestinal infections genital tract see genital infections (and sexually transmitted disease) host susceptibility 1045 infectivity 106 notiable 59, 60 outbreaks and epidemics see epidemics pathogenicity 106 public health physicians in 152 reservoirs/vehicles of infection 1034, 11011 transmission 104, 108 virulence 106 inuenza deaths 18 vaccine 1245 information see health information institutions, variations in incidence within single 20 intention to treat analysis 47, 84 intercensal estimates 52 International Classication of Diseases 245 International Conference on Health Promotion 97 Internet, data sources 51 interobserver variation 28 intervention studies (clinical trials) 13, 4550, 845 allocation in 46 analysis, sequential 47 ethical issues 47 examples 4850 follow-up 47 methods 45 outcome 467 ionizing radiation 1312 IPV (Salk) vaccine 121 ischaemic heart disease see heart disease L labelling effect 139 laboratory-diagnosed infections, data 61 lady almoners 146 Lalonde Report (Canada) 96 lead pollution 128 lead time bias in screening programmes 139 learning disabilities, local authority responsibilities 159 legionnaires disease 9 length bias in screening programmes 139 leukaemia and radiation exposure 36, 44, 132 life expectancy 723 limb malformations, thalidomide associated 34 Lind, James 93 lipid-lowering therapy 87

187

Index
local authority services 154, 1589 logistic regression 41 London smog 129 Lunacy Act (1890) 1467 lunatic asylums 1467 lung cancer 1634 mortalities 346 smoking and see smoking lymphoma and radiation exposure 44 M malaria 126 malignancy see cancer Malthus, Thomas 70 mammography 135 manpower 1601 mass screening 135 maternal issues see mothers (in pregnancy) Maternity and Child Welfare Act (1918) 147 measles 1617, 105, 109, 121 measles/mumps/rubella (MMR) vaccination 117, 1213 medical audit 1756 Medical Ofcers of Health 96, 150 medical profession, history 145 Medical Research Council 87 Medical Research Council treatment trial in mild hypertension 48 Medical Research Council Vitamin Study 50 Mediplus 59 melanoma 167 meningococcus vaccine 107, 123, 126 mental handicap, local authority responsibilities 159 mental health services 149, 1578 mental illness 159, 16970 seasonality of birth and 18 mercury pollution 130 mesothelioma 9, 94 meta-analyses 86 methyl isocyanate gas pollution 132 Midwives Act (1902) 147 migrant populations 23 morbidity statistics 645 source 5861 mortalities (rates) 645, 71, 734 accidental 1702 age-specic 73 cancer see specic site children see abortion; infants; perinatal deaths ethnicity and 22 fall/decline in 745 gender and 212 geographical variations 1920, 21 heart disease see heart disease infectious diseases 745 registration data 536 smoking see smoking social class and 223 standardization see standardization of rates mothers (in pregnancy) health service development 147 occupations, perinatal deaths and 44 multiphasic screening 135 multiple sclerosis 23 multistage sampling 27 multivariate analysis 41 mumps 1212 MMR vaccine 117, 1213 municipal hospitals 146 N National Childhood Encephalopathy Study 120 National Health Insurance Act (1911) 147 National Health Service 15361 changes, 1970s and 1980s 1556 cost 159 early problems 155 efciency 176 local authorities 1589 management 1567 origins 1534 planning 1601 Primary Care Trusts 157 Strategic Health Authorities 157 National Health Service Act (1946) 148 National Health Service trusts 1578 National Institute for Clinical Excellence 158, 182 National Screening Committee 134 National Service Framework for CHD 169 needs see health needs Neisseria meningitides vaccine 123 neonatal death, denition 567 see also perinatal deaths neural tube defects prevention, Medical Research Council Vitamin Study 50 neuroses 18 non-malecence 102 North Karelia Community Trial 98 notiable diseases 59, 60 Nuisances Removal Act (1846) 91, 149 number needed to treat 85 numerator data, errors and bias in 289, 66 nutrition health promotion 100 susceptibility to infection and 105 see also diet O observer variation 289 occupation 223, 25 hazards 60, 132, 172 maternal, perinatal deaths and 44 see also industry odds ratio 41 Ofce of National Statistics 53 operative mortality 64 opportunistic screening 1356 opportunity cost 177

188

Index
oral contraceptives 101 breast cancer and 10 venous thromboembolic disease and 78 ordinal variables 63 osteoarthritis 24 Ottawa Charter 97 Our Healthier Nation 97, 151, 163, 169, 171 ozone layer, destruction 132 P parotitis, epidemic 1212 pathogenicity 106 PE and Sports Programme 101 Peckham Pioneer Health Centre project 96 perinatal deaths 567, 7981 maternal occupation and 44 risk factors 801 periodic changes in incidence 1618 person see individual(s) personal health services 143, 1489 history 1445 personal skills, development 97 pertussis 11920 pertussis (whooping cough) immunization 434, 11920 DTP vaccination 117, 11820 pharmacoepidemiology 5 phenylketonuria screening 133, 135 phocomelia 34 physicians see doctors/physicians PICO 83 pill scare 78 place (in epidemiology) 1920 interactions with other epidemiological factors 23 planning, health service 15 pneumococcus vaccine 125 polio vaccination 116, 118, 1201 pollution 12832 Poor Law (1598) 144 Poor Law Commission 147 Poor Law inrmaries 146 population-dose response 9 populations estimates, in censuses 52 growth rates 702 migrant 23 postcensal 52 projections 52 see also demography poverty and ill health 74, 128 pregnancy see abortion; contraception; mothers (in pregnancy) prevalence (rates) 645 surveys 25 prevention (of disease) primary 95 principles 934 secondary 95 tertiary 95 see also specic problems preventive medicine 149 primary care 148 Primary Care Trusts 156, 157, 173 primary case, denition 107 propagated epidemics 1089 psychological disorders see mental illness public and environmental health services 12732 Public Health Act (1848) 14950 Public Health Act (1871) 150 Public Health Act (1875) 150 Public Health in England 152 public health services 1434, 14952 Q qualitative variables 63 quality framework for health care 173, 174 quality (of health care) 175 quantitative variables 63 questionnaires, reproducibility 301 R rabies vaccine 126 race/ethnicity 22 radiation, ionizing 1312 radon gas 1634 rain, acid 130 random allocation in clinical trials 45, 46, 84 random sampling 26 rates 645 birth see births errors in see errors (in surveys) fertility see fertility (rates) incidence see incidence (rates) mortality see mortalities (rates) prevalence see prevalence (rates) standardization 678 records attributes, absence of 25 unsuitable format 25 Registrar Generals Ofce 53 rehabilitation programmes 101 relative risk 41, 85 reliability of test result 301 Relief of the Poor Act 144 repeatability of test result 301 replication of test 30 reproducibility of test result 301 reproduction see contraception; fertility (rates) research in epidemiology and public health 152 Resource Allocation Working Party 155 respiratory infections, seasonality 17 retrospective studies see casecontrol studies rickets prevention in Asian children in Glasgow 489 risk 8, 11 road trafc accidents 1702 Royal Society for the Prevention of Accidents 98 rubella 1223 MMR vaccine 117, 1213 vaccination 57 rural areas, mortality 201

189

Index
S Sabin vaccine 121 Salk vaccine 121 Salmonella napoli food poisoning 11011 Salmonella typhimurium food poisoning 108, 111 samples (and sampling) 268 bias 278 cluster 27 multistage 27 random 26 stratied 27 systematic 267 sanitary reform movement 149 sanitation, infection and 106 scarlet fever 75 schizophrenia 18 Scientic Advisory Committee on Nutrition (SACN) 100, 182 screening 945, 1338, 1359 accessibility of service 138 antenatal, termination following 138 bias in programmes 139 breast 165 cervix 167 colorectal 165 cost 134, 138 criteria 1368 effectiveness 135 ethics 134, 139 genetic 133 interval for repeat 1389 mass 135 multiphasic 135 opportunistic 1356 phenylketonuria 133, 135 types 1356 search strategies, evidence-based medicine 823 seasonality in incidence 1718 secondary attack rate 108 secondary care 148 secondary case, denition 107 Secretary of State for Health, responsibilities 156 secular trends, incidence 1516 sequential analysis 47 serial interval 107 severe acute respiratory syndrome (SARS) 107 sewage disposal 131 sex see gender sexually transmitted disease see genital infections (and sexually transmitted disease) skin cancer 167 smallpox deaths 75 vaccine 75, 126, 150 smog, London 129 smoking 93, 1689 advice to stop, clinical trial 4950 health promotion 99 mortalities 346, 163 personal responsibility and 98 Snow, John 10, 93 Social Care Department 1589 Social Inequality and Health 5 socio-economic group/social class 223, 128 ischaemic heart disease and 367 Some Department of Health in England 182 specicity (of test) 10, 31 spot-maps 20 staff/personnel, health service 1601 standardization of rates 21, 23, 638 calculation 678 direct 68 indirect 678 standardized mortality ratio 23 Staphylococcus aureus infection 134 Stillbirth (Denition) Act (1992) 56 stillbirths 22, 53, 567, 64 stomach cancer 23 Strategic Health Authorities 156, 157 stratied allocation 46 stratied sample 27 Streptococcus pneumoniae 125 stroke 48, 1689 studies/surveys, epidemiological cross-sectional 25 methods, problems/limitations 2431 sampling 268 types 1113 subject see individual(s) subunit vaccines 116 sudden infant death syndrome 22, 57 suicide 16970 sulphur dioxide pollution 128 supportive environments, creation of 97 surgeons 145 systematic reviews 856 systematic sampling 267 T targeting health status 16272 target populations 45 tertiary care 148 test group, allocation to 46 testicular cancer 1656 tetanus vaccination 119 DTP vaccination 117, 11820 thalidomide 34, 10, 61 time (in epidemiology) 1518 at risk/exposure to causal agent 30, 33 Toronto, Healthy City strategy 151 toxoids 114, 116 transmission (of infectious disease) 104 sexual see genital infections (and sexually transmitted disease) travellers, vaccination 1256 tuberculosis (M. tuberculosis) 94, 105, 146 BCG vaccination 15, 114, 116, 1234 mortalities 1516, 74

190

Index
Type 1 error 85 Type 2 error 85 typhoid, vaccine 125 U UK National Screening Committee 182 urban areas, mortality 201 V vaccination 11526 safety and efcacy 11718 sites 11617 for travellers 1256 types 11516 see also specic vaccines validity 31 variables 63 confounding 78 variation, observer 289 venous thrombosis 7 Virchows triad 7 virulence of pathogen 106 Vitamin Study 50 voluntary hospitals 1456 W waste disposal 131 water pollution 1301 weather, pollution and 129 West Nile fever 107 WHO 151 denitions, of health 50 Health for All by the Year 2000 see Health for All by the Year 2000 immunization targets 124 targets 162 whooping cough 11920 widowhood and fertility 77 workplace see occupation World Health Organization see WHO World Health Organization Healthy Cities 182 World Wide Web, data sources 51 Y Yellow fever vaccine 126

191